delete_reg: don't vacuum or checkpoint

This gets run every time a trace thread is destroyed, which can be once
per target. (That could probably be more efficient, but nonetheless it
can be expected to happen multiple times in one port(1) command.)
Well-behaved API clients will call mportshutdown just before exiting,
which will cause these things to happen once as intended.

(cherry picked from commit 804340d)

src/registry2.0/registry.c

@@ -146,14 +146,6 @@ static void delete_reg(ClientData reg, Tcl_Interp* interp) {
             fprintf(stderr, "reg_optimize: %s\n", error.description);
             reg_error_destruct(&error);
         }
-        if (Tcl_GetAssocData(interp, "registry::needs_vacuum", NULL) != NULL) {
-            reg_vacuum(Tcl_GetAssocData(interp, "registry::db_path", NULL));
-            Tcl_DeleteAssocData(interp, "registry::needs_vacuum");
-        }
-        if (reg_checkpoint((reg_registry*)reg, &error) == 0) {
-            fprintf(stderr, "reg_checkpoint: %s\n", error.description);
-            reg_error_destruct(&error);
-        }
         if (!registry_tcl_detach(interp, (reg_registry*)reg, &error)) {
             fprintf(stderr, "registry_tcl_detach: %s", error.description);
             reg_error_destruct(&error);