Increase the [CPFP outputs values]/[pre-signed transactions feerate]

[darosior]

After #43, i think it'd create a virtuous circle and avoid too much pollution of the utxo set.

[darosior]

By how much ? To your sheds, there is YA bike!

[darosior]

Modified the title, as i also think we ought to be pragmatic and not defer the entire and constant fee bumping to the watchtowers.

In fact, the rationale for increasing the CPFP outputs value implies that we should do this if we want to lower the burden on re-filling watchtowers' funds.

[darosior]

Based on https://github.com/darosior/historical_bitcoin_fees, i believe picking 32sat/vbyte is a reasonable expectation for the median feerate for the year to come.

• We can expect the unvault transaction to be ~300vbytes (taking some breathing room from an estimation with 5 stakeholders). This is a median expected fee of 9600 sats. In order for it to pay for its input (~85 vbytes for 3 managers), and the rest of the transaction (68 vbytes for the normal wallet input, 31vB for the P2WPKH wallet output, 10.5 overhead) it's 194.5 * 32 = 6224 sats new fees needed. This brings us to 15824 sats, or 2.2€ at 1BTC=14000€. We double it in order to not create a dust output which would prevent us to CPFP in the future without a new source of income, and increase the probability that we have more such outputs than actually needed. I therefore propose to use 30_000 sats as the CPFP output value (~4.2€ currently at 1BTC=14000€) for the Unvault Transaction.
• Because of its flexibility, we cannot make a reliable guess on the spend transaction size. Therefore i propose to simply use 2 * 32 * <spend transaction vbytes> as the CPFP output value at its creation. Since the signing process for the Spend Transaction is interactive and lead by the manager creating it, i don't think it would be problem at all. It's also easily verifiable by the watchtowers.

Of course, they are default values and could be made configurable to adapt to the usecase of each business (eg if one is relying on super fast confirmations, they'll likely want to increase it).

[darosior]

Still based on https://github.com/darosior/historical_bitcoin_fees, i believe 85sat/vbyte is a reasonable pick for the revaulting transactions, in order to lighten the charge on each watchtower's wallet during normal operations. It's approximatively 3€ for a 5-managers cancel transaction and 1BTC=14000€, for a transaction which will barely ever be used (not even talking about the emergency one).