Build mkgmap/mkgmap #1221

surrogatePair.patch by Ticker Berkin (slightly modified) "handles UTF16 surrogate pairs so that mkgmap style filters substring and highway-shield give correct results, the LBL trimming leaves the correct amount and neither end up with 1/2 a character. I've looked through all the substring operations and I don't think any others will cause problems. There might just be a few str.length() and str.charAt() or other indexing that might need attention but this would require a lot more searching. I've left the handling for MALFORMED_INPUT so these shouldn't matter."