validate_all() suggestion: decouple validation from fixes

[tm-jc-nacpil]

Decouple applying fixes to validation.

While doing the UAT, the behavior of validate_all() was confusing for me. It seems like when apply_fix = True, the validation columns of the returned dataframe describe the input geometries, not the output. So the immediate interpretation is that the output dataframe still has these errors, even if the geometries are already fixed. I think it would be more useful if it described the output instead, so that a user can safely assume that the dataframe accurately describes the geometries that were returned.

Another suggestion would be to make the default be apply_fix = False, so that a user must intentionally set it to True to modify the geometries.

It would also be nice to have validation be a separate method from actually fixing these errors. There might be usecases where I want to know the errors first, before applying any fixes. The workflow I imagine would be something like

1. validate_all(gdf) to identify errors in the dataframe
2. apply_fixes(gdf, list of validatators to attempt fixing)
3. (optionally) we can also do our own manipulations outside GW to fix these errors or do extra processing
4. validate_all(gdf) to see if the fixes work
5. export the dataframe to whichever format