Add the Result::getColumnName method

[stof]

Q	A
Type	feature
Fixed issues	#5980

Summary

This method allows introspecting the shape of results by knowing the name of columns (combined with the columnCount which already exists).

[derrabus]

Thank you. I think the portability middleware needs a special implementation of that new method.

[stof]

@derrabus indeed, the Portability result needed to handle the case conversion when using its case conversion feature. I added support for it. And

[derrabus]

Thank you! I've opened #6452 as a follow-up.

[morozov]

Thank you @stof.

This is a great addition to the API which should unblock some improvements to the result caching. Specifically, the following issues now can be addressed:

1. If the result contains no rows, its cached version won't return the number of columns.
2. If the result contains columns with the same name (e.g. SELECT a.id, b.id FROM a, b), it won't be cached properly.

Note that the just implemented ArrayResult::getColumnName() is prone to a similar issue: if the result contains no rows, the original result will return column names but the cached version will throw an exception for any column number.

In order to address these issues, instead of accepting list<array<string, mixed>> $data, the ArrayResult constructor could accept (list<string> $columnNames, list<list<mixed>> $data).