Invariant for lb=AS numbers

[eggrobin]

See UTC-177-C46. Fix #534.
See unicode-org/properties#126.

Currently fails as expected:

# Digits are lb=AS in scripts with brahmic line breaking.

[\p{GC=Nd} & $BrahmicLineBreaking] ⊆ \p{LB=AS}

**** START Test Failure 1 ****
## Expected empty, got: 40 [\u1B50-\u1B59\uA9D0-\uA9D9\uAA50-\uAA59\U00011950-\U00011959]

## In [\p{GC=Nd} & $BrahmicLineBreaking]

## But Not In \p{LB=AS}

1B50..1B59	#	[10]	(᭐..᭙)	BALINESE DIGIT ZERO..BALINESE DIGIT NINE
A9D0..A9D9	#	[10]	(꧐..꧙)	JAVANESE DIGIT ZERO..JAVANESE DIGIT NINE
AA50..AA59	#	[10]	(꩐..꩙)	CHAM DIGIT ZERO..CHAM DIGIT NINE
11950..11959	#	[10]	(𑥐..𑥙)	DIVES AKURU DIGIT ZERO..DIVES AKURU DIGIT NINE
**** END Test Failure 1 ****

[markusicu]

ready to take out of draft and merge?

[eggrobin]

ready to take out of draft and merge?

No, waiting on UTC-177 decision, see https://github.com/unicode-org/properties/issues/126.

I’ll merge it once Peter says we have a consensus :-)

[markusicu]

I added a new "wait-for-UTC" label.