When a single commit changes more than 300 changed files than the request is paged. The github module will now handle these.

That would have left the old file in the connected collection dangling.
Now every renamed file will be deleted.
As another edge-case where a file is _added_ and then renamed would try to remove an non-existing file, `github:incremental-delete`
will now treat a not found resource as a successful deletion.

Ignored files are expath-pkg.xml, repo.xml and all .xconf files.
Files that would be ignored was already reported by the dry-run but not for the actual run.
This allows for clients to act on that information and can, for instance, report that a full installation is necessary or that the index configuration has changed.

When determining the changeset for incremental uploads but the local commit hash cannot be found on the remote
the error github:commit-not-found is raised.

- added case is now handled separately as the old entry must be filtered from the del sequence
- get rid of github:remove-or-ignore
  by always filtering lists
  This should improve readability.

An existing file can be removed and then added by **renaming** another file to the previously removed one.
This will now be handled gracefully.