fix: SQL Targets ignore collation when evaluating column data types

[BuzzCutNorman]

This is an attempt to fix #1125

---

📚 Documentation preview 📚: https://meltano-sdk--1385.org.readthedocs.build/en/1385/

[codecov]

Codecov Report

Merging #1385 (fd0548b) into main (9e2ae5c) will increase coverage by 0.00%.
The diff coverage is 85.71%.

@@           Coverage Diff           @@
##             main    #1385   +/-   ##
=======================================
  Coverage   85.19%   85.19%           
=======================================
  Files          54       54           
  Lines        4708     4722   +14     
  Branches      800      803    +3     
=======================================
+ Hits         4011     4023   +12     
- Misses        506      507    +1     
- Partials      191      192    +1     

Impacted Files	Coverage Δ
singer_sdk/connectors/sql.py	78.27% <85.71%> (+0.41%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[edgarrmondragon]

@BuzzCutNorman remove_collation and update_collation both mutate the type object in-place, i.e. you'll notice the returned value points to the same object as the argument:

new_type = remove_collation(current_type)
assert new_type is current_type

So, do you think it makes to only return the collation from remove_collation and return nothing from update_collation. Similar to operations on a dictionary:

>>> d = {1: 2, 2: 3, 3: 4}
>>> d.pop(1)  # Return the removed value
2
>>> d.update(a=0)  # Return nothing, mutate in-place
>>> d
{2: 3, 3: 4, 'a': 0}

[BuzzCutNorman]

So, do you think it makes to only return the collation from remove_collation and return nothing from update_collation. Similar to operations on a dictionary:

@edgarrmondragon, Yes it does make sense and it looks much cleaner.

[edgarrmondragon]

Thanks @BuzzCutNorman, this is g2g!

[BuzzCutNorman]

Thank you @edgarrmondragon ! Couldn't have done it without your excellent guidance. I learned a lot today.