Refactor ByteGroupValueBuilder to use MaybeNullBufferBuilder

[alamb]

Which issue does this PR close?

Follow on to #12269 from @jayzhan211 and #12623
Part of #12680

Rationale for this change

#12623 added a more performant way to track nulls in the PrimitiveGroupValueBuilder. Let's use the same thing for Strings/Binary views to reduce replication

What changes are included in this PR?

1. Refactor ByteGroupValueBuilder to use MaybeNullBufferBuilder to track nulls

Are these changes tested?

Yes, by existing CI

Are there any user-facing changes?

No, this is an internal refactor

Performance (basically all changes within measurement threshold in my opinion)

--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query        ┃  main_base ┃ alamb_boolean_string_groups ┃       Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 0     │     2.19ms │                      2.47ms │ 1.12x slower │
│ QQuery 1     │    38.50ms │                     39.33ms │    no change │
│ QQuery 2     │    94.50ms │                     99.33ms │ 1.05x slower │
│ QQuery 3     │    99.87ms │                     99.86ms │    no change │
│ QQuery 4     │   932.23ms │                    931.81ms │    no change │
│ QQuery 5     │   970.52ms │                    989.52ms │    no change │
│ QQuery 6     │    37.22ms │                     36.55ms │    no change │
│ QQuery 7     │    42.96ms │                     42.71ms │    no change │
│ QQuery 8     │  1371.71ms │                   1354.54ms │    no change │
│ QQuery 9     │  1362.68ms │                   1360.57ms │    no change │
│ QQuery 10    │   342.18ms │                    344.68ms │    no change │
│ QQuery 11    │   395.77ms │                    395.40ms │    no change │
│ QQuery 12    │  1114.55ms │                   1086.39ms │    no change │
│ QQuery 13    │  1592.74ms │                   1714.17ms │ 1.08x slower │
│ QQuery 14    │  1242.36ms │                   1247.77ms │    no change │
│ QQuery 15    │  1087.20ms │                   1095.63ms │    no change │
│ QQuery 16    │  2570.55ms │                   2499.46ms │    no change │
│ QQuery 17    │  2351.69ms │                   2298.15ms │    no change │
│ QQuery 18    │  5035.87ms │                   4970.81ms │    no change │
│ QQuery 19    │    94.64ms │                     96.48ms │    no change │
│ QQuery 20    │  1744.71ms │                   1788.44ms │    no change │
│ QQuery 21    │  2006.06ms │                   2047.56ms │    no change │
│ QQuery 22    │  5202.56ms │                   5173.18ms │    no change │
│ QQuery 23    │ 10514.35ms │                  10580.71ms │    no change │
│ QQuery 24    │   600.74ms │                    609.46ms │    no change │
│ QQuery 25    │   493.64ms │                    516.33ms │    no change │
│ QQuery 26    │   666.07ms │                    666.23ms │    no change │
│ QQuery 27    │  2604.39ms │                   2585.89ms │    no change │
│ QQuery 28    │ 14826.24ms │                  14614.83ms │    no change │
│ QQuery 29    │   526.18ms │                    533.71ms │    no change │
│ QQuery 30    │  1044.06ms │                   1064.09ms │    no change │
│ QQuery 31    │  1137.63ms │                   1123.56ms │    no change │
│ QQuery 32    │  4267.90ms │                   4404.10ms │    no change │
│ QQuery 33    │  5117.07ms │                   5175.47ms │    no change │
│ QQuery 34    │  5179.28ms │                   5224.15ms │    no change │
│ QQuery 35    │  1907.07ms │                   1909.05ms │    no change │
│ QQuery 36    │   280.77ms │                    276.40ms │    no change │
│ QQuery 37    │   123.32ms │                    122.66ms │    no change │
│ QQuery 38    │   147.87ms │                    145.55ms │    no change │
│ QQuery 39    │   788.15ms │                    795.73ms │    no change │
│ QQuery 40    │    56.02ms │                     61.31ms │ 1.09x slower │
│ QQuery 41    │    47.95ms │                     50.08ms │    no change │
│ QQuery 42    │    62.18ms │                     66.27ms │ 1.07x slower │
└──────────────┴────────────┴─────────────────────────────┴──────────────┘

--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query        ┃ main_base ┃ alamb_boolean_string_groups ┃        Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1     │  190.34ms │                    189.37ms │     no change │
│ QQuery 2     │  123.05ms │                    120.34ms │     no change │
│ QQuery 3     │  129.96ms │                    121.71ms │ +1.07x faster │
│ QQuery 4     │   86.80ms │                     89.76ms │     no change │
│ QQuery 5     │  157.98ms │                    165.57ms │     no change │
│ QQuery 6     │   45.93ms │                     45.36ms │     no change │
│ QQuery 7     │  217.67ms │                    192.33ms │ +1.13x faster │
│ QQuery 8     │  176.63ms │                    148.94ms │ +1.19x faster │
│ QQuery 9     │  243.65ms │                    240.19ms │     no change │
│ QQuery 10    │  235.80ms │                    229.71ms │     no change │
│ QQuery 11    │   94.14ms │                     94.43ms │     no change │
│ QQuery 12    │  129.70ms │                    129.90ms │     no change │
│ QQuery 13    │  220.74ms │                    218.86ms │     no change │
│ QQuery 14    │   73.18ms │                     72.92ms │     no change │
│ QQuery 15    │  124.76ms │                    121.70ms │     no change │
│ QQuery 16    │   78.78ms │                     80.14ms │     no change │
│ QQuery 17    │  203.80ms │                    227.06ms │  1.11x slower │
│ QQuery 18    │  329.27ms │                    334.30ms │     no change │
│ QQuery 19    │  126.43ms │                    133.05ms │  1.05x slower │
│ QQuery 20    │  130.58ms │                    137.04ms │     no change │
│ QQuery 21    │  281.02ms │                    271.38ms │     no change │
│ QQuery 22    │   67.55ms │                     65.00ms │     no change │
└──────────────┴───────────┴─────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                          ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (main_base)                     │ 3467.77ms │
│ Total Time (alamb_boolean_string_groups)   │ 3429.05ms │
│ Average Time (main_base)                   │  157.63ms │
│ Average Time (alamb_boolean_string_groups) │  155.87ms │
│ Queries Faster                             │         3 │
│ Queries Slower                             │         2 │
│ Queries with No Change                     │        17 │
└────────────────────────────────────────────┴───────────┘

[Dandandan]

I left a comment here: 36a2003#r1781444375 , I think the code could be simplified a bit by using generics a bit more.

[alamb]

I left a comment here: 36a2003#r1781444375 , I think the code could be simplified a bit by using generics a bit more.

Thank you @Dandandan -- added to list on #12680

(note that link doesn't seem to work for me, but I found the relevant comment #12623 (comment))

[alamb]

This appears to have made another significant improvement in TPCH (presumably because the group keys aren't ever null)

[Dandandan]

👌

[Dandandan]

Really nice performance improvements 🚀

[Dandandan]

Hm actually looking at the performance change of TPCH sf 1,this doesn't seem correct to me (main seems out of date?).

[alamb]

Hm actually looking at the performance change of TPCH sf 1,this doesn't seem correct to me (main seems out of date?).

Yes, sorry -- I am rerunning the performance -- I think this is reporting based on main before @jayzhan211 's changes. Will update shortly

[alamb]

I re-ran benchmarks and posted them above -- TLDR is I don't think there are any significant differences in performance

[alamb]

I can potentially try the const NULLABLE thing too if we think that is worthwhile (I think it is likely that the special case for knowing about nullability is not super applicable to the real world -- aka anything other than TPCH)