fix(fbprophet): Fix weekly frequencies

[john-bodley]

SUMMARY

The fbprophet packages uses pandas.date_range per here for defining the forecast window where the frequency is defined via an offset alias.

Previously the mapping was wrong, i.e., these weeks which use different days of the weeks as anchors (alongside the direction of the anchor, i.e., starting vs. ending) previously all mapped to W which is the same as W-SUN and thus weekly forecasts did not adhere to the correct periodicity defined in the fitted data.

The fix is merely to use the anchored offsets. Pandas doesn't do a great job in specifying whether it's the week starting or week ending on Sunday say for W-SUN, however it's irrelevant for our needs, i.e., for both "Week starting Sunday" and "Week ending Sunday" the data provided by Superset to fbprohet has been bucketed appropriately and anchored on a Sunday and thus all fbprohet needs to know is Sundays (W-SUN) defines the weekly periodicity.

BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF

TESTING INSTRUCTIONS

Tested locally and updated unit tests.

ADDITIONAL INFORMATION

• Has associated issue:
• Required feature flags:
• Changes UI
• Includes DB Migration (follow approval process in SIP-59)
  • Migration is atomic, supports rollback & is backwards-compatible
  • Confirm DB migration upgrade and downgrade tested
  • Runtime estimates and downtime expectations provided
• Introduces new feature or API
• Removes existing feature or API

[codecov]

Codecov Report

Merging #20326 (34043c4) into master (fd12987) will increase coverage by 0.36%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #20326      +/-   ##
==========================================
+ Coverage   66.56%   66.92%   +0.36%     
==========================================
  Files        1738     1738              
  Lines       65025    65597     +572     
  Branches     6883     6883              
==========================================
+ Hits        43281    43902     +621     
+ Misses      19994    19945      -49     
  Partials     1750     1750              

Flag	Coverage Δ
hive	54.20% <ø> (?)
javascript	51.64% <ø> (ø)
mysql	82.51% <ø> (+0.22%)	⬆️
postgres	82.58% <ø> (+0.22%)	⬆️
presto	54.08% <ø> (+0.51%)	⬆️
python	83.02% <ø> (+0.46%)	⬆️
sqlite	82.32% <ø> (+0.23%)	⬆️
unit	50.32% <ø> (+0.17%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
superset/utils/pandas_postprocessing/utils.py	96.55% <ø> (ø)
superset/connectors/sqla/utils.py	90.62% <0.00%> (-0.63%)	⬇️
superset/utils/core.py	90.14% <0.00%> (+0.11%)	⬆️
superset/db_engine_specs/base.py	88.42% <0.00%> (+0.33%)	⬆️
superset/db_engine_specs/databricks.py	92.30% <0.00%> (+1.39%)	⬆️
superset/datasets/api.py	90.49% <0.00%> (+1.96%)	⬆️
superset/connectors/sqla/models.py	92.21% <0.00%> (+2.16%)	⬆️
superset/db_engine_specs/hive.py	85.87% <0.00%> (+15.64%)	⬆️
superset/db_engines/hive.py	85.18% <0.00%> (+85.18%)	⬆️

---

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 fd12987...34043c4. Read the comment docs.

[john-bodley]

@zhaoyongjie would you mind reviewing this PR?

[villebro]

LGTM, thanks for the fix!

[zhaoyongjie]

Tested in my local, goods great. Thanks for the fix. I believed that the master branch code coverage was 100%, and you can safely merge this.