Add t1ha

[leventov]

https://github.com/leo-yuriev/t1ha

[erthink]

This is a problem, since java does not have 64-bit unsigned integers. Therefore wide multiplication (64x64 -> 128) will be unacceptably slow in java implementation.

[leventov]

There is Math.multiplyHigh(long, long) which is supposed to map onto a single x86 instruction since Java 10: https://bugs.openjdk.java.net/browse/JDK-8187684

[erthink]

It is not enough, anyway something like _mymum() is needed.

The proper way is provide _umul128() for Java.