Remove second \tl_to_lowercase:n from l3tl-analysis

This could probably be done using \char_generate:nn but that
would (currently) be problematic in XeTeX if the char was
from outside of the ASCII range (unlikely but not impossible).
Another one to revisting at some stage in the future.

l3experimental/l3str/l3tl-analysis.dtx

@@ -607,7 +607,7 @@
 \cs_new_protected:Npn \@@_a_group:nw #1
   {
     \tex_lccode:D \c_zero = \@@_extract_charcode: \scan_stop:
-    \tl_to_lowercase:n { \tex_toks:D \l_@@_index_int {#1} }
+    \tex_lowercase:D { \tex_toks:D \l_@@_index_int {#1} }
     \if_int_compare:w \tex_lccode:D \c_zero = \tex_escapechar:D
       \int_set:Nn \tex_escapechar:D { 139 - \tex_escapechar:D }
     \fi: