Improve performance for custom business days (GH6584)

[bjonen]

closes #6584

Tests are passing. Incrementing datetime with custom dateoffset is at about twice the speed now. This could probably be optimized more.

[jreback]

can you add a vbench . look for agood place in the vb_suite dir

[bjonen]

Sure, might take a few days though. I'll let you know when I had the time to look into this.

[bjonen]

I've added a small vbench to vbsuite/timeseries.py.

Btw, what is the standard way to run vbsuite on the last commit? I did

%run test_perf.py -t 6305492393 -b 391ed32 -r timeseries

timeseries_custom_bday_incr | 0.0157 | 0.0353 | 0.4437 |

Perhaps it would be good to add some hints on how to use vbench, which script to run etc.

[jreback]

https://github.com/pydata/pandas/wiki/Performance-Testing

[jreback]

pls add a release notes entry under enhancements, referencing #6584

[bjonen]

I've added the release note and rebased. Do you want me to squash this into one commit?

Also, I have started to implement pandas.offsets.CustomBusinessMonth. Should I open a separate PR for this or keep going here?

[jreback]

pls squash

open a new PR/issue for different CustomBusinessMonth

[bjonen]

Ok done

[jreback]

thanks!