[ci] CI canary [WIP/RFC]

[johnkerl]

Internal-only/debug-only canary to examine CI issues arising simultaneously with #1691, while apparently not caused by #1691.

Our post-merge CI has fail-fast set to false -- run all tests to completion -- while our pre-merge CI has fail-fast set to true. This is, generally, a good idea as a large chunk of our CI compute resources are absorbed by pre-merge CI. However, on #1691 I want to see if the Python 3.7 fail there is related only to 3.7. As is on #1691, with fail-fast set to true, the 3.7 fail is cancelling signal from the other CI jobs.

Parent context: #866.

[codecov-commenter]

Codecov Report

Patch has no changes to coverable lines.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

📢 Thoughts on this report? Let us know!.

[johnkerl]

This is progress: 9a08fcc

Before: a CI error involving

ImportError: dlopen(/Users/runner/work/TileDB-SOMA/TileDB-SOMA/apis/python/src/tiledbsoma/pytiledbsoma.cpython-310-darwin.so, 0x0002): symbol not found in flat namespace (__ZN10tiledbsoma8LOG_INFOERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE)

where the symbol demangles to

_tiledbsoma::LOG_INFO(
    std::__cxx11::basic_string<
        char,
        std::char_traits<char>,
        std::allocator<char>
    > const&
)

-- note the placement of the const keyword after the string bit, and right before the &. We do have another error; but I believe this has fixed one of the two issues.

cc @ihnorton @Shelnutt2

[johnkerl]

Superseded by #1695