Fix broken dedup and remove redundant db_spec logic

[villebro]

Fixes regression introduced by PR #4724 that broke column name deduplication. Also remove column name lowercasing logic from db_engine_spec that was made redundant by PR #5178.

[mistercrunch]

@villebro I'm trying to understand how #5178 handles lowercasing or removes the need for it

[villebro]

@mistercrunch The lowercasing hack was introduced to deal with SQLA engines that returned different cases for column names in the cursor.description (db api 1) which was used by e.g. SQL Lab and ResultProxy.keys() (db api 2), which was used for retrieving data. #5178 replaced all v2 api calls with v1 calls, ensuring that the column names are always of the same case.

[codecov-io]

Codecov Report

Merging #5467 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #5467      +/-   ##
==========================================
- Coverage   59.12%   59.11%   -0.01%     
==========================================
  Files         372      372              
  Lines       23762    23751      -11     
  Branches     2758     2758              
==========================================
- Hits        14049    14041       -8     
+ Misses       9698     9695       -3     
  Partials       15       15

Impacted Files	Coverage Δ
superset/db_engine_specs.py	53.25% <ø> (-0.34%)	⬇️
superset/dataframe.py	94.49% <100%> (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 971e9f0...b65ec27. Read the comment docs.

[villebro]

To add context, #4994 was the PR that added the lowercasing logic. Without this new fix reverting the hack, I believe Redshift, Oracle and Snowflake will not work via SQL Lab as of #5178 .

[mistercrunch]

LGTM