Improve point match output

[RussTorres]

Currently the PointMatchClientOutputSchema only produces the name of the output pointmatch collection and the total number of pointmatches in that collection after processing. In common workflows, pointmatching is done in a distributed manner (e.g. per-section montage pointmatching into a single montage correspondence collection). This means that the number of pointmatches is not particularly indicative of anything other than the order in which processing occurred.
A suggestion is to include information about the number of sucessfully matched tiles as well as the number of input pairs from the pair file.

[RussTorres]

Additionally, there could be a fail condition flag for 0 or a certain percentage of expected pairs unmatched that leads to an unsuccessful run from the python side (r.t. setting inliers in the spark input)