[decompiler] More inline vector functions (#3861)

This adds more recognition for inlined vector functions to the
decompiler, which can clean up a bunch of ugly looking code/`rlet`s.


![image](https://github.com/user-attachments/assets/1f7b4627-81bd-481b-b828-76b9f7ba13b3)

Unfortunately, this changes the numbering of ops in the decomp, since
all the vector instructions get combined in a single "operation" by the
decompiler. I really tried to avoid having this ever happen in the
decompiler and this is one of the few cases where it has. So I had to
update a bunch of type casts.

For that reason I haven't turned this on in Jak 2 yet, although I am
planning to do that at some point. (probably at the same time as porting
back a bunch of jak 3 improvements to jak 2)

---------

Co-authored-by: water111 <awaterford1111445@gmail.com>

decompiler/IR2/AtomicOp.cpp

@@ -369,8 +369,12 @@ std::string get_simple_expression_op_name(SimpleExpression::Kind kind) {
       return "vec4dot";
     case SimpleExpression::Kind::VECTOR_LENGTH:
       return "veclength";
+    case SimpleExpression::Kind::VECTOR_LENGTH_SQUARED:
+      return "veclength-squared";
     case SimpleExpression::Kind::VECTOR_PLUS_FLOAT_TIMES:
       return "vecplusfloattimes";
+    case SimpleExpression::Kind::VECTOR_PLUS_TIMES:
+      return "vecplustimes";
     case SimpleExpression::Kind::SET_ON_LESS_THAN:
     case SimpleExpression::Kind::SET_ON_LESS_THAN_IMM:
       return "set-on-less-than";
@@ -440,8 +444,10 @@ int get_simple_expression_arg_count(SimpleExpression::Kind kind) {
     case SimpleExpression::Kind::SET_ON_LESS_THAN_IMM:
       return 2;
     case SimpleExpression::Kind::VECTOR_LENGTH:
+    case SimpleExpression::Kind::VECTOR_LENGTH_SQUARED:
       return 1;
     case SimpleExpression::Kind::VECTOR_PLUS_FLOAT_TIMES:
+    case SimpleExpression::Kind::VECTOR_PLUS_TIMES:
       return 4;
     default:
       ASSERT(false);