Ensure that channels match expected value in conda-build

[jaimergp]

Description

Following up on conda/conda-build#5037. Closes #363.

This issue is caused by a discrepancy in how each PackageRecord is built. conda will take defaults (a multichannel) as the "brand" channel for a PackageRecord, but libmamba takes the actual physical channel (e.g. pkgs/main) as the "brand" channel.

This is not an issue in conda itself because it uses a different mechanism to fetch the PackageRecords, but in conda-build it ends up raising a KeyError because of the channel identifier mismatch.

Let's see if mapping them for conda-build helps.

Checklist - did you ...

• Add a file to the news directory (using the template) for the next release's release notes?
• Add / update necessary tests?
• Add / update outdated documentation?

[jaimergp]

Intriguing that conda-libmamba-solver-src/tests/test_downstream.py::test_build_recipe[jedi] fails in all platforms but Linux-conda-forge with:

WARNING: failed to get install actions, retrying.  exception was: Unsatisfiable dependencies for platform linux-64: {MatchSpec("3.10.*==*_cp310"), MatchSpec("jedi==0.17.2=py310_1")}
Encountered problems while solving:
  - nothing provides python_abi 3.10.* *_cp310 needed by jedi-0.17.2-py310_1

Could not solve for environment specs
The following package could not be installed
└─ jedi 0.17.2.* py310_1 is not installable because it requires
   └─ python_abi 3.10.* *_cp310, which does not exist (perhaps a missing channel).
WARNING:conda_build.environ:failed to get install actions, retrying.  exception was: Unsatisfiable dependencies for platform linux-64: {MatchSpec("3.10.*==*_cp310"), MatchSpec("jedi==0.17.2=py310_1")}
Encountered problems while solving:
  - nothing provides python_abi 3.10.* *_cp310 needed by jedi-0.17.2-py310_1

Could not solve for environment specs
The following package could not be installed
└─ jedi 0.17.2.* py310_1 is not installable because it requires
   └─ python_abi 3.10.* *_cp310, which does not exist (perhaps a missing channel).

Of course, it passes locally if run isolated.

It's also intriguing that newly added test required three reruns to pass, consistently:

../conda-libmamba-solver-src/tests/test_channels.py::test_channel_matchspec PASSED [  1%]
../conda-libmamba-solver-src/tests/test_channels.py::test_channels_prefixdata PASSED [  2%]
../conda-libmamba-solver-src/tests/test_channels.py::test_channels_installed_unavailable PASSED [  3%]
../conda-libmamba-solver-src/tests/test_channels.py::test_mirrors_do_not_leak_channels[_setup_channels_alias] PASSED [  4%]
../conda-libmamba-solver-src/tests/test_channels.py::test_mirrors_do_not_leak_channels[_setup_channels_custom] PASSED [  5%]
../conda-libmamba-solver-src/tests/test_channels.py::test_jax_and_jaxlib PASSED [  6%]
../conda-libmamba-solver-src/tests/test_channels.py::test_encoding_file_paths PASSED [  7%]
../conda-libmamba-solver-src/tests/test_channels.py::test_conda_build_with_aliased_channels RERUN [  8%]  <<<
../conda-libmamba-solver-src/tests/test_channels.py::test_conda_build_with_aliased_channels RERUN [  8%]  <<<
../conda-libmamba-solver-src/tests/test_channels.py::test_conda_build_with_aliased_channels PASSED [  8%]  <<< HERE!
../conda-libmamba-solver-src/tests/test_downstream.py::test_build_recipe[baddeps] PASSED [  9%]
../conda-libmamba-solver-src/tests/test_downstream.py::test_build_recipe[jedi] RERUN [ 10%]
../conda-libmamba-solver-src/tests/test_downstream.py::test_build_recipe[jedi] RERUN [ 10%]
../conda-libmamba-solver-src/tests/test_downstream.py::test_build_recipe[jedi] RERUN [ 10%]
../conda-libmamba-solver-src/tests/test_downstream.py::test_build_recipe[jedi] FAILED [ 10%]

[jezdez]

pre-commit.ci autofix

[jaimergp]

Errors (test_conda_downgrade) are known, and hopefully tamed by #378. So I'd say we merge :)