Save currently executing lua_State in g->cur_L.

This is only a good approximation due to deficiencies in the design of
the Lua/C API. It indicates _some_ valid state that is/was executing.
Also reorder L->cframe stores to achieve a synchronously consistent state.

src/lj_gc.c

@@ -696,7 +696,7 @@ void LJ_FASTCALL lj_gc_step_fixtop(lua_State *L)
 /* Perform multiple GC steps. Called from JIT-compiled code. */
 int LJ_FASTCALL lj_gc_step_jit(global_State *g, MSize steps)
 {
-  lua_State *L = gco2th(gcref(g->jit_L));
+  lua_State *L = gco2th(gcref(g->cur_L));
   L->base = tvref(G(L)->jit_base);
   L->top = curr_topL(L);
   while (steps-- > 0 && lj_gc_step(L) == 0)