fix(ci): compare new benchmarks with base

[Sahil-4555]

Issue:
The benchmark bot was not comparing the new NodeProveMulti/... and NodeProve/... benchmarks with base. The base run was using the older benchmark test harness, so base-lib.txt did not contain those benchmark names.

Fix:
This PR updates the benchmark workflow to sync benchmark test files from the PR checkout into the base checkout before running base benchmarks, only when those files differ. This lets the base run include the same benchmark names, so the bot can show proper comparisons without doing unnecessary sync work.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.75%. Comparing base (4f7b42e) to head (de8a699).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #165   +/-   ##
=======================================
  Coverage   95.75%   95.75%           
=======================================
  Files          47       47           
  Lines       10921    10921           
=======================================
  Hits        10457    10457           
  Misses        290      290           
  Partials      174      174           

Components	Coverage Δ
dynssz	99.61% <ø> (ø)
dynsszgen	92.21% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

Benchmark Results

Library Benchmarks

                                                        │ base-lib.txt │             pr-lib.txt             │
                                                        │    sec/op    │    sec/op     vs base              │
¹ need >= 6 samples for confidence interval at level 0.95
geomean                                                   179.3n         179.5n        +0.12%

                                                        │ base-lib.txt  │              pr-lib.txt               │
                                                        │     B/op      │     B/op       vs base                │
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal
³ summaries must be >0 to compute geomean
geomean                                                               ³                  +0.04%               ³

                                                        │ base-lib.txt │             pr-lib.txt              │
                                                        │  allocs/op   │  allocs/op   vs base                │
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal
³ summaries must be >0 to compute geomean
geomean                                                              ³                +0.00%               ³

Performance Benchmarks

                                          │ base-perf.txt │            pr-perf.txt             │
                                          │    sec/op     │    sec/op     vs base              │
Codegen_BlockMinimal/Unmarshal-4             56.37µ ± ∞ ¹   55.58µ ± ∞ ¹  -1.40% (p=0.008 n=5)
Codegen_BlockMinimal/UnmarshalReader-4       76.49µ ± ∞ ¹   75.48µ ± ∞ ¹  -1.32% (p=0.008 n=5)
Codegen_StateMinimal/UnmarshalReader-4       9.058m ± ∞ ¹   8.871m ± ∞ ¹  -2.07% (p=0.032 n=5)
Reflection_BlockMainnet/Unmarshal-4          56.58µ ± ∞ ¹   55.75µ ± ∞ ¹  -1.47% (p=0.032 n=5)
Reflection_StateMainnet/MarshalWriter-4      5.288m ± ∞ ¹   5.043m ± ∞ ¹  -4.63% (p=0.008 n=5)
Reflection_BlockMinimal/Unmarshal-4          56.80µ ± ∞ ¹   55.87µ ± ∞ ¹  -1.64% (p=0.008 n=5)
Reflection_BlockMinimal/UnmarshalReader-4    77.38µ ± ∞ ¹   75.88µ ± ∞ ¹  -1.94% (p=0.032 n=5)
Reflection_BlockMinimal/MarshalWriter-4      23.09µ ± ∞ ¹   23.31µ ± ∞ ¹  +0.93% (p=0.016 n=5)
Reflection_StateMinimal/MarshalWriter-4      4.565m ± ∞ ¹   4.470m ± ∞ ¹  -2.10% (p=0.008 n=5)
¹ need >= 6 samples for confidence interval at level 0.95
geomean                                      728.6µ         724.5µ        -0.56%

                                          │ base-perf.txt │           pr-perf.txt            │
                                          │     B/op      │     B/op       vs base           │
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal
³ summaries must be >0 to compute geomean
geomean                                                 ³                  ?               ³

                                          │ base-perf.txt │             pr-perf.txt              │
                                          │   allocs/op   │  allocs/op    vs base                │
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal
³ summaries must be >0 to compute geomean
geomean                                                 ³                 +0.00%               ³