test that the generated clients can be run with dependencies installed from conda

[tswast]

I'd like a test session that verifies that the client libraries can be installed via conda. The pandas community strongly encourages conda for installation, so it's important that our packages can be installed into a conda environment (even if it's with pip).

Conda is used to ensure shared C++ libraries such as libprotobuf don't need to be installed more than once (unlike wheels, which distribute their own copies). As a uniform build environment is required (for binary compatibility of compiled libraries), which is defined at https://github.com/conda-forge/conda-forge-pinning-feedstock

Blocked by https://github.com/googleapis/python-bigquery-pandas/issues/398, where I plan to test this for a single library first.

Additional context

• https://github.com/conda-forge/conda-forge-pinning-feedstock/pull/2049 merged on 2021-12-09 which bumped libprotobuf to 3.19. Migration will take time to complete as all packages that depend on libprotobuf need to be rebuilt.
• Conda installation problems with Python 3.10 https://github.com/conda-forge/google-cloud-bigquery-feedstock/issues/106, https://github.com/conda-forge/tensorflow-feedstock/pull/189 This was caused due to protobuf global pin to 3.18.1 and a Python 3.10 build wasn't provided for the Python protobuf package 3.18.1 https://github.com/conda-forge/protobuf-feedstock/issues/131 Even with that, it was partially exasperated by https://github.com/googleapis/proto-plus-python/pull/270/files#r765932065 which bumped the minimum protobuf library in proto-plus.