[ENH]: Optimize GetCollections and remove usage of raw gorm (#5274)

## Description of changes

An earlier [change](https://github.com/chroma-core/chroma/pull/5268)
attempted to use a CTE to optimize the GetCollections query in the event
where the databases pkey is fully specified. This led to using raw gorm
which led to a few unexpected binding issues when converting from gorm
sql to raw sql, leading us to
[revert](https://github.com/chroma-core/chroma/commit/71aae948da980954d89994132def214a27ccfcaa)
that change here. In this change, we achieve the same goal of optimizing
GetCollections but avoid the use of CTE's entirely. We get the required
database_id from databases within a predicate subquery. If the predicate
subquery has a LIMIT 1 in it, Postgres is known to treat this as a CTE.
Since gorm plays much more nicely with subqueries than CTE's this is a
much better fix.

- Improvements & Bug fixes
  - ...
- New functionality
  - ...

## Test plan

_How are these changes tested?_

- [ ] Tests pass locally with `pytest` for python, `yarn test` for js,
`cargo test` for rust

## Migration plan

_Are there any migrations, or any forwards/backwards compatibility
changes needed in order to make sure this change deploys reliably?_

## Observability plan

_What is the plan to instrument and monitor this change?_

## Documentation Changes

_Are all docstrings for user-facing APIs updated if required? Do we need
to make documentation changes in the [docs
section](https://github.com/chroma-core/chroma/tree/main/docs/docs.trychroma.com)?_

---------

Co-authored-by: propel-code-bot[bot] <203372662+propel-code-bot[bot]@users.noreply.github.com>