Change S0 calculation to enforce Again < Hard < Good < Easy.py

[Expertium]

Relevant comment: ankitects/anki#3367 (comment)

These changes will prevent situations where Again = Hard, or Hard = Good, or anything like that. But we still need to do something about the other problem. Read my comment above.

[L-M-Sherlock]

This change may induce a problem: if the initial stability of again is very large (perhaps due to the lack of data, the user rarely presses again in the first learning), the initial stabilities of hard, good and easy will be enforced correctly.

The problem you want to solve has been resolved in open-spaced-repetition/fsrs-rs#212

[Expertium]

Can you at least keep the clampings I proposed?
Also, smooth_and_fill isn't strict enough. We need Again < Hard < Good < Easy, not just Again <= Hard <= Good <= Easy

[L-M-Sherlock]

I mean the clamps can induce the problem. So I cannot keep it.

[Expertium]

Alright, close this PR, but please come up with a way to enforce strict monotonicty, aka "less than", not "less than or equal to"

[L-M-Sherlock]

OK, I will consider it. Could you open a new issue to record this feature?