many: use Mu::Regex instead of std::regex

The former is PCRE-compatible, and faster than std::regex.

lib/utils/mu-utils.cc

@@ -39,7 +39,6 @@
 #include <cinttypes>
 #include <charconv>
 #include <limits>
-#include <regex>
 
 #include <glib.h>
 #include <glib/gprintf.h>
@@ -261,15 +260,6 @@ Mu::split(const std::string& str, char sepa)
 	return vec;
 }
 
-std::vector<std::string>
-Mu::split(const std::string& str, const std::regex& sepa_rx)
-{
-	std::sregex_token_iterator it(str.begin(), str.end(), sepa_rx, -1);
-	std::sregex_token_iterator end;
-
-	return {it, end};
-}
-
 std::string
 Mu::join(const std::vector<std::string>& svec, const std::string& sepa)
 {

lib/utils/mu-utils.hh

@@ -33,7 +33,6 @@
 #include <type_traits>
 #include <algorithm>
 #include <numeric>
-#include <regex>
 
 #include "mu-utils-format.hh"
 #include "mu-option.hh"
@@ -97,16 +96,6 @@ std::vector<std::string> split(const std::string& str, const std::string& sepa);
  */
 std::vector<std::string> split(const std::string& str, char sepa);
 
-/**
- * Split a string in parts
- *
- * @param str a string
- * @param sepa the separator regex
- *
- * @return the parts.
- */
-std::vector<std::string> split(const std::string& str, const std::regex& sepa_rx);
-
 /**
  * Join the strings in svec into a string, separated by sepa
  *
@@ -173,8 +162,6 @@ bool locale_workaround();
  */
 bool timezone_available(const std::string& tz);
 
-
-
 /**
  * Well-known runtime paths
  *

lib/utils/tests/test-utils.cc

@@ -220,9 +220,6 @@ test_split()
 	// char sepa
 	assert_equal_svec(split("axbxc", 'x'), {"a", "b", "c"});
 	assert_equal_svec(split("axbxcx", 'x'), {"a", "b", "c", ""});
-
-	// rx sexp
-	assert_equal_svec(split("axbyc", std::regex("[xy]")), {"a", "b", "c"});
 }
 
 static void