lnwallet+rpcserver: fix weight calculation for taproot channels

[yyforyongyu]

Fix the weight calculation found in multiple places.

Fix #8036

[guggero]

Thanks for the quick fix!

[guggero]

Very nice! Can confirm the itest catches the issue that is fixed in the PR. LGTM 🎉

[Roasbeef]

To answer the question above re why this results in a lower fee rate in the sanity check instead of a higher one:

• We use the correct weight values everywhere else up to this point, so we estimate against a "fully loaded" commitment tx (witness, etc).
• When we get down to this check, we actually over estimate the weight since we use the segwit v0 witness value. For the musig2 channels, the witness is ~4x smaller, so by using the old constant we over estimate the weight for the sanity check, which leads to a lower fee rate.

[Roasbeef]

LGTM 👙