fix : include datatype in memo key in sanitization function

[vignesh14052002]

• closes BUG: Datatypes not preserved on pd.read_excel #60088
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

[jbrockmendel]

Perf impact? This is here for a reason.

[vignesh14052002]

This is the commit that introduces the change
6c31cab

I dont't understand why this was added, but reverting solves the below issue

from pandas._libs.parsers import sanitize_objects

values = np.array([1,"NA",True],dtype=object)
print("Values before sanitization:",values)
sanitize_objects(values,na_values={"NA"})
print("Values after sanitization:",values)

output

Values before sanitization: [1 'NA' True]
Values after sanitization: [1 nan 1]

Eventhough the sanitization parts works fine (NA->nan), it is converting True to 1 and that is due to the memo
I have some issues setting up the environment to run performance tests

[jbrockmendel]

I dont't understand why this was added, but reverting solves the below issue

The commit message was "memoize objects when reading from file to reduce memory footprint". So removing it will likely balloon memory footprint. Instead of removing it, might be more effective to just check for 0, 1, True, False explicitly and let other values be memoized?

[vignesh14052002]

Thanks, now i understand about memory footprint. skipping memoization just for those 4 values might not be a good approach, because what if the data contains only mixture of those 4 values? it can blew up the memory

I have included type of the value too in memo key, which will solve this