Tweak OSR heuristic to fix gaussian-blur regression

My recent change https://crrev.com/c/v8/v8/+/2698057 changed the size of
bytecode for most functions, and attempted to update other heuristic
values to match. However, it caused V8 to be slightly too eager to
perform on-stack replacement in JetStream 2's gaussian-blur test case,
so that the function got compiled separately for each of two nested
loops rather than just once for the outer loop. This is the smallest
change that restores the previous behavior in that benchmark.

Bug: chromium:1179571
Change-Id: I03e98d6bff7355b775c1fdaf495e7444e7c6f095
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2704882Reviewed-by: Mythri Alle <mythria@chromium.org>
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#72849}

src/execution/runtime-profiler.cc

@@ -32,7 +32,7 @@ static const int kBytecodeSizeAllowancePerTick = 1100;
 // Maximum size in bytes of generate code for a function to allow OSR.
 static const int kOSRBytecodeSizeAllowanceBase = 119;
 
-static const int kOSRBytecodeSizeAllowancePerTick = 43;
+static const int kOSRBytecodeSizeAllowancePerTick = 44;
 
 // Maximum size in bytes of generated code for a function to be optimized
 // the very first time it is seen on the stack.