Slightly change score strategy for columns

[hiddenmemory]

Previously, if a column had no match, it would short circuit returning None, stopping subsequent columns from potentially matching.

This change means that non-matching columns are scored as zero, allowing subsequent columns to be matched against.

If the match has zero score after all columns, then we return None, retaining previous behaviour.

All existing tests pass.

[pascalkuthe]

this is not how columns are intended to be used. It's an AND match not an OR match. For example in helix diagnostic picker %severity ERR %message type will show you all type errors. With this change that would not work anymore.

[alexrutar]

Would it be worth permitting more general (user-provided) column score strategies? Something like making the MultiPattern::score function generic over a "score strategy" trait.

This would handle the comment in the codebase about potentially allowing column weights.

Then, in principle, as long as the "score strategy" trait is sufficiently general, someone could implement the behaviour suggested in this PR if desired.

[pascalkuthe]

I would be more option to that but I think the better way to do this would be to add general OR predicate (the one aspect of FZF we haven't ported yet) and then we could have a general mechanism to include OR patterns that would also apply to columns (like colum1: foo & column2: bar | colum3: baz).