Support for handling broken excel formulas as strings

[skytwosea]

Currently, when excel sheets are read into a dataframe using pd.read_excel(file, engine="calamine",), all cells that contain broken formulas are transcribed as Null/None, and there is no option to set this behaviour to capture the Excel error code(s). Reading a sheet with dtype='string' does not address this issue: it enforces the string type for Excel cells with valid data, but broken formulas are still read in as Null/None.

This is not ideal: information is being lost. Many sheets have valid empty/null cells that should be distinguished from cells with broken formulas via an error code (#NULL!, #REF!, #VALUE!, etc).

Calamine provides functions to deserialize types as <T> or None: see here. fastexcel should provide a parameter or config switch to set this behaviour.

[skytwosea]

edit: I should have looked harder... duplicate of #229