[GHA] fix several codecov issues

[kiukchung]

This PR does a few things:

1. Excludes keras_core/legacy from coverage metrics since this module there for BC reasons
2. Adds the test backend (jax, tensorflow, torch, numpy) to flags to allow tracking coverage metrics per backend. Note that each backend's coverage will be lower than the aggregate (e.g. keras_core will have higher coverage than keras_core-jax), so this flag should be used to track trends rather than impose thresholds.
3. Moves coverage and pytest configuration from setup.cfg to pyproject.toml to keep all project configs in toml vs cfg (unfortuantely flake8 configs need to stay in setup.cfg since flake8 doesn't support pyproject.toml yet: pyproject.toml (PEP 518) support PyCQA/flake8#234)
4. Adds fail_ci_if_error: true to codecov upload steps to fail the CI (and force the author to retry the workflow) on transient codecov upload failures of the form

    Unable to locate build via Github Actions API. Please upload with the Codecov repository upload token to resolve issue.', code='not_found
   

   This happens when we hit codecov API rate limits since we don't use CODECOV_TOKEN (see "Unable to locate build via Github Actions API" for the public repository codecov/feedback#126). We could circumvent the issue by adding the token as a secret but GH secrets are not visible from forks, implying that contributors will not be able to get coverage diffs for their PRs. To allow this, the only option seems to be to hardcode the token (which seems ok since the token can be setup with limited permissions) but hardcoding tokens is just sketchy...

[kiukchung]

will graduate from draft commit after I validate GHA and check codecov reports
codecov metrics + flags look good

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: +6.79% 🎉

Comparison is base (a465816) 76.82% compared to head (742e71c) 83.62%.
Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     codecov/codecov-action#912      +/-   ##
==========================================
+ Coverage   76.82%   83.62%   +6.79%     
==========================================
  Files         329      318      -11     
  Lines       31427    28390    -3037     
  Branches     6112     5409     -703     
==========================================
- Hits        24144    23740     -404     
+ Misses       5719     3150    -2569     
+ Partials     1564     1500      -64     

Flag	Coverage Δ
keras_core	83.51% <ø> (+6.78%)	⬆️
keras_core-jax	67.58% <ø> (?)
keras_core-numpy	60.73% <ø> (?)
keras_core-tensorflow	67.07% <ø> (?)
keras_core-torch	69.42% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

see 13 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fchollet]

LGTM -- thank you!