Benchmarks improvements

[grarco]

We should try to improve the benchmarks (if possible) with the followings:

• Review the benchmarks for IBC transactions (we've added new features since the benches were written)
• Split the benchmark of verify_shielded_tx in the masp VP into separate benches for check_spend, check_convert, check_output and final_check so that we can better track gas at runtime depending on the number of notes in the transaction
• Refactor run::tx and run::vp to avoid benchmarking the call to fetch_and_compile (which is already tracked separately)