Fix: handle column default types correctly in SchemaCache

[mursidx]

This pull request addresses issue #3706 by improving the handling of column default values in the SchemaCache.
The previous implementation did not correctly account for certain data types, which could lead to incorrect metadata retrieval.
This fix ensures that all relevant cases are covered, including scenarios where ad.adbin is not NULL.

Additionally, an entry has been added to the CHANGELOG.md for this fix.

Fixes #3706

[laurenceisla]

Thanks for the PR. Here are my observations:

[laurenceisla]

This still gives the same result since select pg_get_expr(null,null) returns null. What we need here is to get the column default before the domain default.

A test is needed to verify that changes work. Something similar to this one, but with another table with a column that overrides the domain default (like the one mentioned in the issue):

postgrest/test/spec/Feature/Query/InsertSpec.hs

Lines 566 to 573 in 5ca969f

	it "inserts a default on a DOMAIN with default" $
	request methodPost "/evil_friends?columns=id,name" [("Prefer", "return=representation"), ("Prefer", "missing=default")]
	[json| { "name": "Lu" } |]
	`shouldRespondWith`
	[json| [{"id": 666, "name": "Lu"}] |]
	{ matchStatus = 201
	, matchHeaders = ["Preference-Applied" <:> "missing=default, return=representation"]
	}

[laurenceisla]

There is already an "Unreleased" section at the beginning of the document. The entry should go under the "Fixed" subtitle. Check the other entries for the format.

[mursidx]

@laurenceisla ,
I have updated the SchemaCache.hs to prioritize column defaults as suggested. I also added a corresponding test case in InsertSpec.hs to verify the functionality and updated the CHANGELOG.md accordingly.

[wolfgangwalther]

There is already a Fixed h3-section under the Unreleased h2-section. It should be added to that section. Also you are still not following the format as the other log entries below. Have a close look where the issue number is placed, note that the author is listed at the end, etc...

[wolfgangwalther]

Laurence asked you to write a test-case similar to this one, but with some changes (i.e. test for column default, not domain default)...

Instead of adjusting the test-case, so that it actually matches what you are trying to achieve.. you just copy-pasted the exact same thing, which is a duplicate of the test-case right after.

What are you trying to achieve?

[wolfgangwalther]

Also note that all tests are failing, because your SQL is invalid + the style check is not passing either.