Network Quantitative Timeliness Agreements, aka QTAs

[nfrisby]

@njd42 suggested these a while ago, and there was some planning around them, but for various reasons (I think personnel churn), the work disappeared.

The basic idea (feel free to edit this, Neil) is that the different components/teams need to specify their expected performance profiles, and add some benchmarks/tests to catch performance regressions/violations.

For now, I'm just opening this same Issue mutatis mutandi in the Plutus, Ledger, Consensus, and Network repositories. Each team will at least need to consider what they want to do and want others to do.