CI: 32 bit maybe_indices_to_slice

[jbrockmendel]

• closes #xxxx
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[jreback]

hmm you are backing this to fix the 32-bit tests? does this have a perf hit?

[jbrockmendel]

Not sure yet, still trying to get this passing

[jbrockmendel]

got some more troubleshooting to do, but in the interim, some greenish: #37365, #37402, #37449, #37451, #37456, #37470, #37472, #37475
and green but less trivial #37425, #37436, #37423, #37425, #37179, #37164, #34683

The deluge is going to slow to a crawl since the next task on my list is a doozy: figuring out why the categorical validators all behave slightly differently.

[jbrockmendel]

alright, got the 32bit working

[jreback]

can u run asvs

[jbrockmendel]

can u run asvs

i can start them running overnight. this should be totally neutral on 64bit since np.intp is an alias for np.int64

[jbrockmendel]

forgot to run asvs overnight, just started them now.

[jbrockmendel]

ran with -b index

BENCHMARKS NOT SIGNIFICANTLY CHANGED.

[simonjayhawkins]

@jreback if this is ready to merge will backport, otherwise can just skip failing 32 bit test on MacPython.

[jreback]

thanks @jbrockmendel

[simonjayhawkins]

@meeseeksdev backport 1.1.x