Merge bitcoin/bitcoin#30352: policy: Add PayToAnchor(P2A), OP_1 <0x4e73> as a standard output script for spending

75648cea5a test: add P2A ProduceSignature coverage (Greg Sanders)
7998ce6b20 Add release note for P2A output feature (Greg Sanders)
71c9b02a04 test: add P2A coverage for decodescript (Greg Sanders)
1349e9ec15 test: Add anchor mempool acceptance test (Greg Sanders)
9d89209937 policy: stop 3rd party wtxid malleability of anchor spend (Greg Sanders)
b60aaf8b23 policy: make anchor spend standard (Greg Sanders)
455fca86cf policy: Add OP_1 <0x4e73> as a standard output type (Greg Sanders)

Pull request description:

  This is a sub-feature taken out of the original proposal for ephemeral anchors #30239

  This PR makes *spending* of `OP_1 <0x4e73>` (i.e. `bc1pfeessrawgf`) standard. Creation of this output type is already standard.

  Any future witness output types are considered relay-standard to create, but not to spend. This preserves upgrade hooks, such as a completely new output type for a softfork such as BIP341.  It also gives us a bit of room to use a new output type for policy uses.

  This particular sized witness program has no other known use-cases (https://bitcoin.stackexchange.com/a/110664/17078), s it affords insufficient cryptographic security for a secure commitment to data, such as a script or a public key. This makes this type of output "keyless", or unauthenticated.

  As a witness program, the `scriptSig` of the input MUST be blank, by BIP141. This helps ensure txid-stability of the spending transaction, which may be required for smart contracting wallets. If we do not use segwit, a miner can simply insert an `OP_NOP` in the `scriptSig` without effecting the result of program execution.

  An additional relay restriction is to disallow non-empty witness data, which an adversary may use to penalize the "honest" transactor when RBF'ing the transaction due to the incremental fee requirement of RBF rules.

  The intended use-case for this output type is to "anchor" the transaction with a spending child to bring exogenous CPFP fees into the transaction package, encouraging the inclusion of the package in a block. The minimal size of creation and spending of this output makes it an attractive contrast to outputs like `p2sh(OP_TRUE)` and `p2wsh(OP_TRUE)` which
  are significantly larger in vbyte terms.

  Combined with TRUC transactions which limits the size of child transactions significantly, this is an attractive option for presigned transactions that need to be fee-bumped after the fact.

ACKs for top commit:
  sdaftuar:
    utACK 75648cea5a
  theStack:
    re-ACK 75648cea5a
  ismaelsadeeq:
    re-ACK 75648cea5a via [diff](e7ce6dc070..75648cea5a)
  glozow:
    ACK 75648cea5a
  tdb3:
    ACK 75648cea5a

Tree-SHA512: d529de23d20857e6cdb40fa611d0446b49989eaafed06c28280e8fd1897f1ed8d89a4eabbec1bbf8df3d319910066c3dbbba5a70a87ff0b2967d5205db32ad1e

src/test/transaction_tests.cpp

@@ -1026,6 +1026,14 @@ BOOST_AUTO_TEST_CASE(test_IsStandard)
         t.vout[0].nValue = 239;
         CheckIsNotStandard(t, "dust");
     }
+
+    // Check anchor outputs
+    t.vout[0].scriptPubKey = CScript() << OP_1 << std::vector<unsigned char>{0x4e, 0x73};
+    BOOST_CHECK(t.vout[0].scriptPubKey.IsPayToAnchor());
+    t.vout[0].nValue = 240;
+    CheckIsStandard(t);
+    t.vout[0].nValue = 239;
+    CheckIsNotStandard(t, "dust");
 }
 
 BOOST_AUTO_TEST_SUITE_END()