Tests: point the guard-bit comments at eeGuardedAddSub

fpuAddSubGuarded is not a name in the tree. The interpreter's adder is
eeGuardedAddSub; fpuGuardedAddSub is the INI bool the recompilers gate
on.

tests/ctest/core/recompilers/ee_fpu_guarded_addsub_console_tests.cpp

@@ -6,7 +6,7 @@
 // The EE adder keeps no guard bits, so a cancelling subtraction (or mixed-sign
 // add) renormalises left over the lost bits and lands one ULP toward zero from
 // the IEEE answer. The mechanism, the mask and the emitters it was ported from
-// are at fpuAddSubGuarded in FPU.cpp. This file is the console pin for the
+// are at eeGuardedAddSub in FPU.cpp. This file is the console pin for the
 // class; EeRecFpuGuardBit (ee_rec_fpu_guardbit_tests.cpp) covers the emitted
 // code and the fpuGuardedAddSub opt-out.
 //
@@ -157,7 +157,7 @@ struct ScopedEeFpEnv
 };
 
 // The reference model, transcribed from x86 FPU_ADD_SUB (iFPU.cpp); the rule it
-// implements is stated at fpuAddSubGuarded in FPU.cpp.
+// implements is stated at eeGuardedAddSub in FPU.cpp.
 //
 // __noinline: several callers below pass compile-time constants, and an inlined
 // body would let the compiler fold the arithmetic at build time, under the
@@ -206,7 +206,7 @@ __noinline u32 ModelPlainAddSub(u32 a, u32 b, bool issub)
 } // namespace
 
 // ---------------------------------------------------------------------------
-// The interpreter failed all thirteen of these before fpuAddSubGuarded().
+// The interpreter failed all thirteen of these before eeGuardedAddSub().
 // ---------------------------------------------------------------------------
 TEST(EeFpuGuardedAddSubConsole, InterpMatchesConsoleOnEveryGuardBitRow)
 {

tests/ctest/core/recompilers/ee_rec_fpu_guardbit_tests.cpp

@@ -16,7 +16,7 @@
 // don't need it. These tests run under the default (on);
 // DisableEmitsPlainOpMatchingInterp at the bottom pins the opt-out path.
 //
-// The interpreter guards too (fpuAddSubGuarded, FPU.cpp), so in the default
+// The interpreter guards too (eeGuardedAddSub, FPU.cpp), so in the default
 // configuration the two engines agree on every guard-sensitive input and the
 // tests below can use Run()'s JIT-vs-interp auto-diff rather than
 // RunJitNoDiff(). They part company only on the opt-out, which the interpreter