Fixes for Pandas 1.1.0

[thehomebrewnerd]

Fixes #1072

• Update wrangle.py to change handling for special date offsets to support both older and newer versions of pandas
• Use dask single-threaded scheduler for compute calls to avoid issue with Dask tests failing with threaded scheduler - see Dask Issue 6454
• Bump pandas minimum version to 1.0.0 as changes to needed in wrangle.py to work with 1.1.0 will cause failures with older versions.
• Bump dask minimum version to 2.12.0 to avoid error that happens with version 2.11.0 which does not implement dd.to_numeric()

[codecov]

Codecov Report

Merging #1079 into main will decrease coverage by 0.01%.
The diff coverage is 94.44%.

@@            Coverage Diff             @@
##             main    #1079      +/-   ##
==========================================
- Coverage   98.36%   98.35%   -0.02%     
==========================================
  Files         126      126              
  Lines       13158    13161       +3     
==========================================
+ Hits        12943    12944       +1     
- Misses        215      217       +2     

Impacted Files	Coverage Δ
featuretools/utils/wrangle.py	75.25% <85.71%> (-0.79%)	⬇️
...mputational_backend/test_feature_set_calculator.py	97.93% <100.00%> (ø)
featuretools/tests/conftest.py	100.00% <100.00%> (ø)
...ools/primitives/standard/aggregation_primitives.py	96.92% <0.00%> (-0.31%)	⬇️

---

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 3569038...2a557e0. Read the comment docs.

[thehomebrewnerd]

The fixes implemented in this PR were verified to work with pandas versions as old as v1.0.0, but they fail on versions older than that.

[thehomebrewnerd]

Based on limited testing, this code chunk_sum = chunk_sum.astype('int64') identified as a codecov issue appears to be hit with pandas 1.0.0 or 1.0.5, but not with pandas 1.1.0.

[rwedge]

For all these scheduler="single-threaded" options, could we set that via an environmental variable or something so we don't need to declare it each compute call

If not, I think we should make a SCHEDULER_TYPE variable and set that to "singlet-threaded", then use SCHEDULER-TYPE everywhere instead

[rwedge]

are there any changes that require using 1.0 or higher?

[thehomebrewnerd]

Yes - the changes in wrangle.py don't work with older versions because of the way the date offsets are handled. There might be a way to come up with something that would work with both, but I'd have to spend some more time looking into it.

[thehomebrewnerd]

Might be able to use a different if-else construct to check for the two scenarios that were present in the previous code. I'll check.

[rwedge]

Looks good