Rename aspects of the codebase which are confusing

[PaliC]

This issue continues the discussion from #128

My proposed changes are

TorchBenchSuite → ModelTracesSuite
datasets → test_suites
huggingface_op_trace → backendbench_tests

Motivation for the renames

TorchBenchSuite → ModelTracesSuite

TorchBench is a misnomer here, as this test suite actually contains models from HuggingFace Transformers, TIMM, and TorchBench (which is actually the smallest contributor iirc). ModelTraces better captures that the test suite consists of model traces used in PyTorch CI. Given that we plan to expand this test suite primarily through additional traces, this name is more appropriate.

datasets → test_suites

This was poor naming on my part. When creating new test suites, I incorrectly called them datasets, which caused confusion.

huggingface_op_trace → backendbench_tests

I updated the repository name where ModelTracesSuite lives to correct the misnomer—it's a repository of test suites, not HuggingFace op traces.

[PaliC]

@msaroufim Gonna put my response to #128 (comment) here (quoted below)

: I'm not a big fan of the new names or the old ones :P. Just so you don't get randomized updating the PR mind discussing these in an issue instead? Personally I'd prefer things that fully explain what something is like HF_TOP_MODEL_SHAPE. Tests is vague it could refer to things in test it could refer to suite? datasets could mean anything

How about we drop test then. I'm not really sure what HF_TOP_MODEL_SHAPE is referring to in the context of the three things I'm renaming.

How about
datasets → bb_suites (as these are literally backendbench suites)

huggingface_op_trace → backendbench_suites (similar as above)