Numerous EWTS->Unicode and especially EWTS->TMW improvements.

Fixed ordering of Unicode wowels.  [ku+A] gives the correct Unicode
now, e.g.

EWTS->TMW looks better for some wacky wowels like, I'm guessing here, [ku+A].

EWTS->TMW should now give errors any time the full input isn't used.
Previously, wacky wowels like [kai+-i] would lead to some droppage.

EWTS->TMW->Unicode testing is now in effect.  This found a ton of
EWTS->TMW bugs, most or all of which are fixed now.

TMW->Unicode is improved/fixed for {
\u5350,\u534D,\u0F88+k,\u0F88+kh,U }.  (Why U?  "\u0f75" is
discouraged in favor of "\u0f71\u0f74".)

NOTE: TMW_RTF_TO_THDL_WYLIETest is still disabled for the nightly
builds' sake, but I ran it in my sandbox and it passed.

source/org/thdl/tib/input/DuffPane.java

@@ -1377,7 +1377,7 @@ public void paste(int offset)
                 if (TibetanMachineWeb.isPunc(val)) { //punctuation
                     val = TibetanMachineWeb.getWylieForPunc(val);
 
-                    if (val.charAt(0) == TibetanMachineWeb.BINDU)
+                    if (val.startsWith(THDLWylieConstants.BINDU))
                         putBindu();
 
                     else {