Use bottom_up in fuzzer template substitution (#4769)

Replaces `top_down` with `bottom_up` to avoid redundant traversals in
replacement terms, improving performance for large terms. While
replacement terms are usually not large, this change enables an
optimization in komet that results in a 3x speedup.

The optimization in komet works by embedding large configuration cells
without variables directly into a replacement term. This prevents the
substitution algorithm from needlessly traversing these large terms,
significantly reducing overhead.
* runtimeverification/komet#69

Co-authored-by: automergerpr-permission-manager[bot] <190534181+automergerpr-permission-manager[bot]@users.noreply.github.com>

Author: bbyalcinkaya

pyk/src/pyk/ktool/kfuzz.py

@@ -169,7 +169,7 @@ def sub(p: Pattern) -> Pattern:
                 return p
 
         handler.handle_test(subst_case)
-        test_pattern = template.top_down(sub)
+        test_pattern = template.bottom_up(sub)
         res = llvm_interpret_raw(definition_dir, test_pattern.text, check=False)
 
         try: