Investigate slow Registry benchmarks

[HCastano]

The Relayer pallet runtime benchmarks can be quite slow to run, especially if not built
in release mode.

I'm convinced that it has to do with the benchmark setup process being done in polynomial
time (nested for-loops when calling add_non_syncing_validators).

for i in 0..SIG_PARTIES {
    let validators = add_non_syncing_validators::<T>(sig_party_size, 0, i as u8);
    <ThresholdToStash<T>>::insert(&threshold_account, &validators[i]);
}

We should look into whether or not there's a way to set the benchmarks up in a more
efficient manner.

[HCastano]

Closing, I don't think anybody has time to look into it. Not a huge issue since we're only running this on releases.