Support for Attach/Detach Chunk

[melihmutlu]

Introduce attach_chunk and detach_chunk procedures to support attaching
and detaching chunks to/from hypertables.

Detaching a chunk preserves non-dimensional check constraints, indexes,
and foreign keys on the chunk itself. However, any foreign keys
referencing the chunk and dimensional check constraints are dropped.

Attaching a chunk preserves existing constraints, indexes, foreign
keys, etc. Additional constraints inherited from the hypertable are
created, including dimensional constraints. Existing indexes on the
chunk are compared against those on the hypertable; if a matching index
is found, it is reused as the chunk’s index.

Both attach_chunk and detach_chunk acquire a ShareUpdateExclusiveLock
on the hypertable and an AccessExclusiveLock on the chunk, similar to
PostgreSQL partitioning.

Compressed chunks are not currently supported, but it's planned for
future version.

[mkindahl]

Some minor comments below, but otherwise looks fine. As you mentioned, you're working on the tests so waiting with the approval until I've seen those as well. You also need to add upgrade and downgrade code.

[codecov]

Codecov Report

Attention: Patch coverage is 82.64463% with 42 lines in your changes missing coverage. Please review.

Project coverage is 82.03%. Comparing base (f0c3113) to head (6d6a4f1).
Report is 36 commits behind head on main.

Files with missing lines	Patch %	Lines
tsl/src/chunk_api.c	70.37%	5 Missing and 19 partials ⚠️
src/chunk_constraint.c	85.52%	1 Missing and 10 partials ⚠️
src/constraint.c	85.71%	0 Missing and 3 partials ⚠️
src/chunk.c	90.47%	0 Missing and 2 partials ⚠️
src/indexing.c	90.47%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #8223      +/-   ##
==========================================
- Coverage   82.23%   82.03%   -0.21%     
==========================================
  Files         257      257              
  Lines       48497    48736     +239     
  Branches    12218    12295      +77     
==========================================
+ Hits        39882    39981      +99     
- Misses       3730     3904     +174     
+ Partials     4885     4851      -34     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[erimatnor]

It would be good to have some isolation tests as well to see how attach/detach interacts with concurrent:

• compression
• queries
• inserts
• other attach/detach operations
• vacuum
  etc.

[mkindahl]

Mostly nits, and a few requests for additional tests

[erimatnor]

Good job with the isolation tests.

I think the PR looks pretty good now but there are some things to fix.

• The attach test ends prematurely due to a constraint violation so that needs to be fixed.
• You need to handle partitioning func a bit differently in constraint checking code
• There were some (potentially) unexpected errors in some isolation test cases that you should double-check

See inline comments.

[erimatnor]

Perhaps you can also expand on the description for the commit before you merge? It's a bit short right now. Perhaps write something about behavior, locking, etc.

[erimatnor]

LGTM. Still have a couple of questions. For example, one about attaching a chunk with a different owner than the hypertable.

[erimatnor]

Superuser can attach regardless of owners of the hypertable and the chunk.

But can the superuser attach a chunk with owner 'foo' while the hypertable has owner 'bar'? If that works, I am not sure we'd like to support it unless we implicitly change the owner of the chunk to align with the hypertable.

[erimatnor]

Note typos in commit message. E.g., "procuders"