upgrade to Unicode 15

[carenas]

To avoid a regression in test4 and test7 where the first character of the set was unassigned and no longer qualifies for R bidi as shown by:

$ ./ucptest -s 590 5c6 200f 10805                                                                                                                   
U+0590 L   Control: Unassigned, unknown, Other, >֐<
U+05C6 R   Punctuation: Other punctuation, hebrew, Other, [alphabetic, caseignorable, extender, graphemebase, idcontinue, idstart, xidcontinue, xidstart], >׆<
U+200F R   Control: Format, common, Control, [dash, defaultignorablecodepoint, deprecated, emojipresentation, joincontrol, lowercase, patternwhitespace, radical, regionalindicator, softdotted, xidcontinue, xidstart], >‏<
U+10805 R   Letter: Other letter, cypriot, Other, [alphabetic, diacritic, graphemebase, idcontinue, xidcontinue], >𐠅<

Had to revert several of the updates to DerivedBidiClass.txt that came with the released files.

The gist of it is that (as described in the comments in that file) the new release relies in setting defaults for several characters that are not identified and that were marked in the previous release as "reserved".

The way the defaults are defined uses a comment that starts with "@missing:" and that we should IMHO parse and add to our lookup logic, but doing so has been punted for now.