Rename optimize_coerce To _in, Add coerce_ni

in cases where a coerce_ni is used on the result of
a coerce_in, we can just use the original integer.

Normally, the combination of these ops would cause
truncation of integers to only ~52 bits of precision,
but for many operations, like elems, it's likely that
the full integer was meant to be used.

This optimization may want to be limited to nqp, or
perhaps limited to the result of certain ops.

src/spesh/optimize.c

@@ -894,7 +894,7 @@ static void optimize_can_op(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshBB *
 }
 
 /* If we have a const_i and a coerce_in, we can emit a const_n instead. */
-static void optimize_coerce(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshBB *bb, MVMSpeshIns *ins) {
+static void optimize_coerce_in(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshBB *bb, MVMSpeshIns *ins) {
     MVMSpeshFacts *facts = MVM_spesh_get_facts(tc, g, ins->operands[1]);
 
     if (facts->flags & MVM_SPESH_FACT_KNOWN_VALUE) {
@@ -912,6 +912,24 @@ static void optimize_coerce(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshBB *
     }
 }
 
+/* A coerce_ni after a coerce_in can become a set. Hooray! */
+static void optimize_coerce_ni(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshBB *bb, MVMSpeshIns *ins) {
+    MVMSpeshFacts *facts = MVM_spesh_get_facts(tc, g, ins->operands[1]);
+
+    if (facts->writer && facts->writer->info->opcode == MVM_OP_coerce_in) {
+        MVMSpeshFacts *result_facts = MVM_spesh_get_facts(tc, g, ins->operands[0]);
+
+        MVM_spesh_usages_delete(tc, g, facts, ins);
+
+        ins->info = MVM_op_get_op(MVM_OP_set);
+        ins->operands[1].reg = facts->writer->operands[1].reg;
+
+        MVM_spesh_usages_add_by_reg(tc, g, ins->operands[1], ins);
+
+        MVM_spesh_graph_add_comment(tc, g, ins, "Removed redundant int-to-num-to-int coerce");
+    }
+}
+
 /* If we know the type of a significant operand, we might try to specialize by
  * representation. */
 static void optimize_repr_op(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshBB *bb,
@@ -2654,7 +2672,10 @@ static void optimize_bb_switch(MVMThreadContext *tc, MVMSpeshGraph *g, MVMSpeshB
             break;
         }
         case MVM_OP_coerce_in:
-            optimize_coerce(tc, g, bb, ins);
+            optimize_coerce_in(tc, g, bb, ins);
+            break;
+        case MVM_OP_coerce_ni:
+            optimize_coerce_ni(tc, g, bb, ins);
             break;
         case MVM_OP_smrt_numify:
         case MVM_OP_smrt_strify: