While going over UTS #39 I noticed that the internalSkeleton function is defined as:

1. Convert X to NFD format, as described in UAX15.
2. Remove any characters in X that have the property Default_Ignorable_Code_Point.
3. Concatenate the prototypes for each character in X according to the specified data, producing a string of exemplar characters.
4. Reapply NFD.

But in the current implementation, we leave out step 2 which probably does not change the behaviour in most cases. Still, it is likely a good idea to fix this.

This pull request adds this step.

Notes:

• I took the liberty to refactor load_properties in unicode.py to effectively remove a copy of the function.
• I regenerated the tables as is. There seem to have been some changes to the IDENTIFIER_TYPE table upstream which are included here in addition to the new table.