fix: add space in no_lost_declarations regex (#14080)

As [reported](#14048 (comment)) and [requested](#14048 (comment)) by Yael!

I think that this change is an improvement, but I would be happier if there was a consensus that the `def`, `theorem`, `structure`,... keywords are followed by a space and not a line break before their identifier.

Author: adomani

scripts/no_lost_declarations.sh

@@ -18,9 +18,9 @@ else
 fi |
   ## purge `@[...]`, to attempt to catch declaration names
   sed 's=@\[[^]]*\] ==; s=noncomputable ==; s=nonrec ==; s=protected ==' |
-  ## extract lines that begin with '[+-]' followed by the input `theorem` or `lemma`
-  ## in the `git diff`
-  awk -v regex="^[+-]${begs}" 'BEGIN{ paired=0; added=0; removed=0 }
+  ## extract lines that begin with '[+-]' followed by the input `theorem`, `lemma`,...
+  ## and then a space in the `git diff`
+  awk -v regex="^[+-]${begs} " 'BEGIN{ paired=0; added=0; removed=0 }
     /^--- a\//    { minusFile=$2 }  ## the path to the old file
     /^\+\+\+ b\// { plusFile=$2 }   ## the path to the new file
     ($0 ~ regex){
@@ -132,5 +132,5 @@ instance (priority := high) {to be a nameless} instance :=
 def testingLongDiff1 im a def
 def testingLongDiff2 im a def
 def testingLongDiff3 im a def
-def testingLongDiff4 im a def
+@[trying to fool you] instance. the messing dot
 ReferenceTest