Lint spaces in qualified names

[jpers36]

Fixes #1754

This rule change adds another type of unnecessary whitespace to L039. Any whitespace or newline segments found within a reference segment type are removed.

Per @jpy-git 's suggestion this has been implemented as a change to an existing rule, but this could be spun out into its own rule if that's preferred.

Anti-pattern:

SELECT [thistable] . [col]
FROM [thisdatabase] .     [thisschema]
   . [thistable]

Pattern:

SELECT [thistable].[col]
FROM [thisdatabase].[thisschema].[thistable]

[codecov]

Codecov Report

Merging #2130 (117c647) into main (e63a17d) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##              main     #2130   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files          148       148           
  Lines        10667     10674    +7     
=========================================
+ Hits         10667     10674    +7     

Impacted Files	Coverage Δ
src/sqlfluff/rules/L039.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e63a17d...117c647. Read the comment docs.

[jpy-git]

One thing that #2137 highlighted is that the changes to this rule are triggering warnings in some situations.

For example, in the unit test in test/rules/std_L003_L036_L039_combo_test.py if you try fixing the query you'll see these warnings:

As far as I can see there's nothing in this that should be touched by your addition but it is. Can you investigate please?

query here:

WITH example AS (
        SELECT my_id,
            other_thing,
            one_more
        FROM
            my_table
    )

    SELECT *
    FROM example

[jpers36]

@jpy-git I looked into it just now, and it's due to #1304 as well. The real fix would be to get that issue fixed. The workaround would be to extend the workaround I added so that it ignores multiple leading whitespace segments and not just the first segment. It shouldn't be hard, but the first solutions that come to mind are all inelegant.

[jpy-git]

@jpers36 Looks good and your suggestion to handle the leading whitespace seems pragmatic 👍

Re: the incorrect parse tree issue causing those leading whitespaces in the object reference, we're well aware of the issue and it's a fairly complicated one. I'm doing some initial investigation atm. We can come back and tidy up the extra logic here in the future once that's fixed 😄

Nice work, LGTM! 🚀