BUG: DTI.searchsorted accepting invalid types/dtypes

[jbrockmendel]

Analogous to #30763 which fixed the same issue for PeriodIndex. After this will be a PR to fix TimedeltaIndex, then one to move the fixes up to the EA methods (and share searchsorted).

[jschendel]

If a plain list is passed should there be an attempt to infer the dtype (or just coerce to an array-like) to determine if the list is valid? E.g. should the following succeed?

In [1]: import pandas as pd; pd.__version__
Out[1]: '0.26.0.dev0+1685.g9565b5666'

In [2]: dti = pd.date_range("2020", periods=3)

In [3]: dti.searchsorted(list(dti))
---------------------------------------------------------------------------
TypeError: searchsorted requires compatible dtype or scalar, not list

[jreback]

If a plain list is passed should there be an attempt to infer the dtype (or just coerce to an array-like) to determine if the list is valid? E.g. should the following succeed?

In [1]: import pandas as pd; pd.__version__
Out[1]: '0.26.0.dev0+1685.g9565b5666'

In [2]: dti = pd.date_range("2020", periods=3)

In [3]: dti.searchsorted(list(dti))
---------------------------------------------------------------------------
TypeError: searchsorted requires compatible dtype or scalar, not list

i suppose we could wrap list-likes with pa.array, but this would be an enhancement; can you open an issue.

[jreback]

thanks @jbrockmendel