Reconstruct matches when reading annotations

[JeltevanBoheemen]

When reading annotations from a corrected excel file, allresults.allmatches are not updated. The following should happen:

• Detect which annotations are not in the corrected file, remove the corresponding matches from allmatches
• Detect new annotations. Reconstruct matches for these using their word position to find the corresponding node in the parsed transcript.

Be aware that this may still result in allresults.allmatches not correctly corresponding to allresults.coreresults and allresults.annotations. This is because some queries match more than one word, and the annotation Excel format is unable to express this. For scoring purposes, assuming the matches have a span of 1 is fine.

[JeltevanBoheemen]

On hold, unknown if necessary.