Improve load times by skipping serialization of entities when unecessary.

[arjo129]

🦟 Bug fix

Fixes #

Summary

I was investigating
gazebosim/gazebo_test_cases#1576 , in my investigation it came to my notice that sdf::Root takes a long time to be constructed. While I am submitting changes upstream to reduce the impact of the creation of sdf::Root, in the event that we don't serialize a model, I'm proposing we just send an empty string accross. This should minimize both network traffic and make load times more manageable.

Checklist

• Signed all commits for DCO
• Added tests
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• While waiting for a review on your PR, please help review another open pull request to support the maintainers

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by messages.

🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸🔸

[azeey]

I remember we added this serializing of sdf::Model to support scaling on the GUI, but that was never implemented. Maybe @caguero remembers more. I'm not sure if we need it for anything else.

[caguero]

I remember we added this serializing of sdf::Model to support scaling on the GUI, but that was never implemented. Maybe @caguero remembers more. I'm not sure if we need it for anything else.

All that I remembered was that it started motivated to support scaling. I'm not sure if after that any other piece is using it.

[arjo129]

cc: @caguero

Should we remove the serialization all together then? The speed up from removing it is quite a bit. 3k_worlds.sdf take well over 300s (5minutes) to load with serialization, if I remove it we can load in 22seconds an almost 13x improvement.

Each model instance takes about 0.1s to deserialize (which is not very good).

I do think we should eventually address the root cause which I've documented here: gazebosim/sdformat#1478

[azeey]

@arjo129 having long load times is not great, but since this has been an issue for a long time and only affects loading large files, I would not consider it a critical bug. I'd rather not make this change while we are in code freeze.

[arjo129]

Sure lets revisit this after the release. That being said, I would say that 3k worlds only loads on the largest computer I have access to (which happens to be a cloud instance). It fails to load on any of my local machines.

This fix also makes loading openrmf demo worlds a lot faster.