fix: import salt.utils.jid explicitly in pgjsonb returner (#69042)

[SAY-5]

What does this PR do?

Restores the explicit import salt.utils.jid in the pgjsonb returner. The module calls salt.utils.jid.gen_jid() in prep_jid() (line 461) and salt.utils.jid.format_jid_instance() in get_jids() (line 436), but currently only imports salt.utils.data and salt.utils.job. The salt.utils.jid reference works today as a side effect of salt.utils.args being loaded transitively via salt.utils.data. Any refactor of that import chain silently breaks pgjsonb with AttributeError: module 'salt.utils' has no attribute 'jid'.

prep_jid() is called from master.py:_prep_jid() on every job publish. That caller only handles KeyError, TypeError so the latent AttributeError would propagate and prevent the master from publishing jobs when master_job_cache: pgjsonb is in use.

A regression test in tests/pytests/unit/returners/test_pgjsonb.py asserts that salt.utils.jid is reachable through the module's globals so future refactors can no longer drop the explicit import.

What issues does this PR fix or reference?

Fixes #69042

Previous Behavior

prep_jid / get_jids worked only because salt.utils.jid was loaded transitively. In an interpreter where it had not been loaded yet, both functions raised AttributeError: module 'salt.utils' has no attribute 'jid'.

New Behavior

The explicit import salt.utils.jid is restored alongside the existing salt.utils.data / salt.utils.job imports, so both call sites resolve regardless of which other modules have been imported beforehand.

Merge requirements satisfied?

• Docs
• Changelog - changelog/69042.fixed.md
• Tests written/updated

Commits signed with GPG?

No

[twangboy]

Please rebase once 3006.x branch tests are green. Fix conflicts if any.
Make this on the earliest branch where the bug exists. Likely 3006.x

[twangboy]

Please rebase and fix failing tests