Character set disagreements

[rossabaker]

From #1:

failing seed for GeneratorsCheck.generates xml is 2vlpy64yIOKuEnqKnWrPi2mnlyiIXmu-8aZUOMf2qcL=
[info] ! GeneratorsCheck.generates xml: Exception raised on property evaluation.
[info] > Exception: java.lang.IllegalArgumentException: ʾ館뭆焲뭓鄹 must be an XML Name
[info] Failed: Total 1, Failed 0, Errors 1, Passed 0
[error] Error during tests:
[error] 	org.typelevel.scalacheck.xml.GeneratorsCheck
[error] (coreNative / Test / test) sbt.TestsFailedException: Tests unsuccessful

We targeted XML 1.0, Fourth Edition, which is what Xerces still uses. scala-xml has its own definitions. That leads to the following discrepancies:

scala> letter.filterNot(scala.xml.Utility.isNameStart)
val res6: Seq[Char] = List(ʻ, ʼ, ʽ, ʾ, ʿ, ˀ, ˁ, ՙ, ۥ, ۦ, ℮)

scala> (letter ++ digit ++ ".-_".toSeq ++ combiningChar ++ extender).filterNot(scala.xml.Utility.isNameChar)
val res15: Seq[Char] = List(℮, ۝, ۞, ·)

We could just use scala-xml's character sets, but I'm not sure it round trips.

[rossabaker]

scala/scala-xml#607

[rossabaker]

#5 fixes this for our purposes. Upstream discussion is linked above, which could let us simplify, but our work here is done for now.