Put an empty ^$ after wblank

I thought we were good because

```sh
$ echo '^ikkje<adv>/ikkje$ «[[tf:i:a]]s\^å[[/]]»' |cg-proc -1ng nob-nno.genprefs.rlx.bin
ikkje «[[tf:i:a]]s\^å[[/]]»
```

worked, but

If there's an analysis after the unmerged word blank, cg-proc errors out:

```sh
$ echo '^ikkje<adv>/ikkje$ «[[tf:i:a]]s\^å» ^.<sent>/.$' |cg-proc -1ng nob-nno.genprefs.rlx.bin
Error: Word-bound blank was not immediately prior to token on line 0
```

Fair enough, so lt-merge must put an empty ^$ after word blanks to
appease cg-proc. Seems like tf-inject finds the right point at which
to end it anyway?

lttoolbox/fst_processor.cc

@@ -1967,6 +1967,7 @@ FSTProcessor::quoteMerge(InputFile& input, UFILE *output)
       if(surface.size() > 0) {
         surface += reader.blank;
         appendEscaped(surface, reader.wblank);
+        if(!reader.wblank.empty()) { appendEscaped(surface, "^$"_u); } // otherwise cg-proc will Error wordblank not prior to token
       }
       else {
         // The initial blank should just be output before the merged LU:

tests/lt_merge/__init__.py

@@ -38,8 +38,8 @@ class EscapeTest(MergeTest):
 
 class WordblankTest(MergeTest):
     # Using r'' to avoid doubling escapes even more:
-    inputs = [r'^«/«<lquot><MERGE_BEG>$[[tf:i:a]]^ve\/ldig/v<adv>$[[/]]^»/»<rquot><MERGE_END>$']
-    expectedOutputs = [r'^«\[\[tf:i:a\]\]ve\\\/ldig\[\[\/\]\]»/«\[\[tf:i:a\]\]ve\\\/ldig\[\[\/\]\]»<MERGED>$']
+    inputs = [r'^«/«<lquot><MERGE_BEG>$[[tf:i:a]]^ve\/ldig/v<adv>$^»/»<rquot><MERGE_END>$']
+    expectedOutputs = [r'^«\[\[tf:i:a\]\]\^\$ve\\\/ldig»/«\[\[tf:i:a\]\]\^\$ve\\\/ldig»<MERGED>$']
 
 
 class SimpleUnmergeTest(MergeTest):
@@ -52,5 +52,5 @@ class SimpleUnmergeTest(MergeTest):
 class EscapedUnmergeTest(MergeTest):
     procflags = ['--unmerge']
     # Using r'' to avoid doubling escapes even more:
-    inputs = [r'^ikkje<adv>/ikkje$ ^«\[\[tf:i:a\]\]s\\\^å\[\[\/\]\]»<MERGED>/«\[\[tf:i:a\]\]s\\\^å\[\[\/\]\]»$']
-    expectedOutputs = [r'^ikkje<adv>/ikkje$ «[[tf:i:a]]s\^å[[/]]»']
+    inputs = [r'^ikkje<adv>/ikkje$ ^«\[\[tf:i:a\]\]\^\$s\\\^å»<MERGED>/«\[\[tf:i:a\]\]\^\$s\\\^å»$']
+    expectedOutputs = [r'^ikkje<adv>/ikkje$ «[[tf:i:a]]^$s\^å»']