Merge supports analysis files

[lobis]

This updates the file merging logic so that metadata are also merged.

It is assumed that this is only used to merge files that are generated from an identical configuration file (up to run number, random seed, etc.). The first file metadata will be propagated to the merge file, except for TRestGeant4Metadata that will be the product of the merging of other metadatas (add up number of primaries, etc.).

Metadata, geometry, event and analysis tree will be present in the final file (product of merge).

Originally this merge macro modified the event id of the events if it was duplicated across the new merge file. This is no longer easy to do since we had to use TFileMerger to merge the trees (could not find another way), so there will be duplicated event ids.

[lobis]

So even though the implementation at the time of writing this comment seems to work okay: event, analysis trees are correctly duplicated and metadata are merged, there are some wierd issues when doing the merge where it fails on large number of events after processing many events. Not sure why but I will not merge this until (if) the problem is addressed...

[jgalan]

I think doing the merging of analysis tree data is in the scope of TRestDataSet.

[lobis]

I think doing the merging of analysis tree data is in the scope of TRestDataSet.

The original idea of this PR was to have some process that would also merge the metadata. If you just need to merge the analysis tree you can just instantiate a RDataFrame with multiple files as source and do an snapshot.