SQL == CSV For YAML Generation

[eloise-nebula]

This branch adds a regression test verifying that the SQL→YAML pipeline produces output equivalent to the existing CSV + meme pipeline, along with the fixes needed to make that comparison hold.

• Added tests/test_db_to_yaml.py::test_compare_yaml, which builds device/area data from both pipelines and asserts they match.
• Added use_meme/suppress_warnings options to YAMLGenerator/YAMLWriter, plus a new directory_service.verify_address(), so PV lookups can run against the local DB instead of live meme calls during tests.
• pmt_metadata was removed because the information is now available in oracle.
• Fixed area filtering and dict key ordering in db_to_yaml.build() and beampath whitespace handling in YAMLGenerator, which were causing mismatches between the two pipelines' output.

Once this PR is merged, slac_db.db_to_yaml.get_device() will be able to fully replace YAML files for all devices.

[williamColocho]

No comments. Looks good to me.

[kabanaty]

Hi Eloise, this all looks good to me. We'll find out quickly after the summer downtime if the PMT metadata is working correctly but as of now I think it will be fine. Thanks for the hard work!