SQLAlchemy 2.0 upgrades (part 2)

[jdavcs]

Follow-up to #16434. Covers ORM usage (ref). This PR is ready for review.

For follow-up PRs:

• lib/galaxy/datatypes
• lib/galaxy/jobs
• lib/galaxy/managers
• lib/galaxy/metadata
• lib/galaxy/tool_shed
• lib/galaxy/web
• lib/galaxy/webapps/galaxy/controllers
• lib/galaxy_test
• lib/tool_shed
• build_initial_query and friends (web/framework/*, webapps/*)
• Fix bug in User.all_roles query + use SA2.0 #16761
• model.DatasetCollection._get_nested_collection_attributes()
• model.DatasetCollection.populated_optimized
• managers.users (tags: convert from Query, apply union correctly)

Ref: #12541

How to test the changes?

(Select all options that apply)

• I've included appropriate automated tests.
• This is a refactoring of components with existing test coverage.
• Instructions for manual testing are as follows:
  1. [add testing steps and prerequisites here if you didn't write automated tests covering all your changes]

License

• I agree to license these and all my past contributions to the core galaxy codebase under the MIT license.

[jmchilton]

I wonder if we should just merge the new tool shed backend before you go making lots of changes to the code - that branch already refactors a bunch of stuff.

[jdavcs]

I wonder if we should just merge the new tool shed backend before you go making lots of changes to the code - that branch already refactors a bunch of stuff.

I would be very much in favor of that; I wasn't sure how close that was to merging. Is it the full #15639 or a subset of it? (I'll then go through it and review (at least) the db/SA-related code).

[jmchilton]

Well we had decided to deploy it and test it first... that is seeming like a mistaken call given how we're not making progress on that.

I already found and commented on one issue with the previous DB commit.

[jmchilton]

Rebase and ping me?

[jdavcs]

@jmchilton: rebased

[jdavcs]

Rebased again - just for the sake of starting a new branch on top of this. The pre-rebase version was all green (except for one unrelated selenium failure).

[jmchilton]

Very nice! Thanks for breaking these up into reviewable pieces.