infra: core-contrib-tests randomly fails

[emdneto]

https://github.com/open-telemetry/opentelemetry-python/actions/runs/13767616206/job/38498115997

Some tests of core_contrib_test.yml workflow are failing randomly with:

Using Python 3.8.18 environment at: .tox/py38-test-instrumentation-httpx-1
   Updating https://github.com/open-telemetry/opentelemetry-python.git (bc8f29f986da0db702f7e04f8f4e9fb811bb5bcb)
  × Failed to download and build `opentelemetry-semantic-conventions @
  │ git+[https://github.com/open-telemetry/opentelemetry-python.git@bc8f29f986da0db702f7e04f8f4e9fb811bb5bcb#egg=opentelemetry-semantic-conventions&subdirectory=opentelemetry-semantic-conventions`](https://github.com/open-telemetry/opentelemetry-python.git@bc8f29f986da0db702f7e04f8f4e9fb811bb5bcb#egg=opentelemetry-semantic-conventions&subdirectory=opentelemetry-semantic-conventions%60)
  ├─▶ Git operation failed
  ├─▶ failed to find branch, tag, or commit
  │   `bc8f29f986da0db702f7e04f8f4e9fb811bb5bcb`
  ╰─▶ process didn't exit successfully: `/usr/bin/git rev-parse
      'bc8f29f986da0db702f7e04f8f4e9fb811bb5bcb^0'` (exit status: 128)
      --- stdout
      bc8f29f986da0db702f7e04f8f4e9fb811bb5bcb^0

      --- stderr
      fatal: ambiguous argument 'bc8f[29](https://github.com/open-telemetry/opentelemetry-python/actions/runs/13767616206/job/38498115997#step:5:30)f986da0db702f7e04f8f4e9fb811bb5bcb^0':
      unknown revision or path not in the working tree.
      Use '--' to separate paths from revisions, like this:
      'git <command> [<revision>...] -- [<file>...]'

  py38-test-instrumentation-httpx-1: FAIL code 1 (3.08 seconds)
  evaluation failed :( (3.53 seconds)

I found some related cases here: actions/runner-images#9632
Investigate that

[jomcgi]

This has been a nightmare for me - I don't have the permissions to retry specific tests when I run into a failure and rerunning all tests increases that chance that I run into this again.

I think we should be accounting for a failure to git pull with some sort of retry logic on the git step - when we're pulling frequently some failures should be expected I think github's SLA is ~99.9% for git operations.

I can put this in a PR for consideration / discussion.