bip-tapscript: merkle key tree isn't exactly a replacement for multisig

[sipa]

Review comment by @gmaxwell:

Even without MuSig, due to the signature hash committing to the tapleaf hash, signers must commit up front to the exact subset that will be signing. That means you can't just give a signature to the other signers and not care about who will complete it. A solution is providing signatures for all matching leaves, but this may be unreasonable for large multisig.

[sipa]

See #109.