patch 8.2.5057: using gettimeofday() for timeout is very inefficient

Problem:    Using gettimeofday() for timeout is very inefficient.
Solution:   Set a platform dependent timer. (Paul Ollis, closes #10505)
diff --git a/src/ex_cmds.c b/src/ex_cmds.c
index e01d067..34a740d 100644
--- a/src/ex_cmds.c
+++ b/src/ex_cmds.c
@@ -4006,7 +4006,7 @@
 		); ++lnum)
     {
 	nmatch = vim_regexec_multi(&regmatch, curwin, curbuf, lnum,
-						       (colnr_T)0, NULL, NULL);
+						       (colnr_T)0, NULL);
 	if (nmatch)
 	{
 	    colnr_T	copycol;
@@ -4663,7 +4663,7 @@
 			|| nmatch_tl > 0
 			|| (nmatch = vim_regexec_multi(&regmatch, curwin,
 							curbuf, sub_firstlnum,
-						    matchcol, NULL, NULL)) == 0
+						    matchcol, NULL)) == 0
 			|| regmatch.startpos[0].lnum > 0)
 		{
 		    if (new_start != NULL)
@@ -4728,7 +4728,7 @@
 		    }
 		    if (nmatch == -1 && !lastone)
 			nmatch = vim_regexec_multi(&regmatch, curwin, curbuf,
-					  sub_firstlnum, matchcol, NULL, NULL);
+					  sub_firstlnum, matchcol, NULL);
 
 		    /*
 		     * 5. break if there isn't another match in this line
@@ -4992,7 +4992,7 @@
     {
 	lnum = curwin->w_cursor.lnum;
 	match = vim_regexec_multi(&regmatch, curwin, curbuf, lnum,
-						       (colnr_T)0, NULL, NULL);
+						       (colnr_T)0, NULL);
 	if ((type == 'g' && match) || (type == 'v' && !match))
 	    global_exe_one(cmd, lnum);
     }
@@ -5005,7 +5005,7 @@
 	{
 	    // a match on this line?
 	    match = vim_regexec_multi(&regmatch, curwin, curbuf, lnum,
-						       (colnr_T)0, NULL, NULL);
+						       (colnr_T)0, NULL);
 	    if (regmatch.regprog == NULL)
 		break;  // re-compiling regprog failed
 	    if ((type == 'g' && match) || (type == 'v' && !match))