[niteshbalusu11]

Closes #1255.

bitcoincore-rpc 0.19 failed to decode getmempoolentry / verbose getrawmempool (missing bip125-replaceable) and getblockstats (missing utxo_size_inc). Both fields are now returned.

• bip125-replaceable on getmempoolentry, verbose getrawmempool, and verbose getmempoolancestors / getmempooldescendants. It is Core's IsRBFOptIn: true when the tx or an in-mempool ancestor has an input sequence at or below 0xfffffffd. It is not hardcoded to true as the issue suggests: Core 32.1 still reports false for an opt-out tx under full RBF (feature_rbf.py test_fullrbf), and full RBF here still replaces it. MempoolHub::signals_bip125 checks the tx and its ancestor set under one read lock. This overlaps the bip125 part of draft rpc: answers Core clients rely on (captaind / bark on rbitcoin) #1191, which has been idle.
• getblockstats utxo_size_inc / utxo_size_inc_actual. Each output adds its serialized CTxOut plus PER_UTXO_OVERHEAD, each spent prevout subtracts the same, and _actual adds only outputs that enter the UTXO set. The overhead is 40, Core v32.1's sizeof(COutPoint) + sizeof(uint32_t). The issue cites Core 31.1, which counted a spent-flag byte (41). This repo pins v32.1. A stamped block reads its input edges in one input.body span read and each parent's outputs once, only when the selected stats include a size stat. The stamped outputs pass also sums total_out now, so Query::non_coinbase_total_out is gone.
• The Core functional proxy no longer computes the two size stats from Esplora. rpc_blockstats.py and its self-test are deleted, and rpc_getblockstats.py runs against the node alone.

Not in this PR: gettxoutsetinfo stays -32601. COMPAT.md lists it as never: the node keeps no UTXO set to hash or count.

Proof: the new assertions fail before each fix and pass after. The mempool journey covers a signaling parent, an opt-out child that inherits true, an opt-out local tx at false, and full RBF replacing that tx. getblockstats pins regtest genesis at 116/0, and the stamped and reconstruct paths both match a tally from getblock verbosity 3 prevouts for a block with an OP_RETURN output. Core 32.1's rpc_getblockstats.py passes locally against the node with the shim removed. My first cut used 41, and this script caught it at height 101 (163/75 instead of 161/74). Local fmt, clippy, deny, ast-grep, ratchets, ci-os-smoke.sh, the harness self-tests, and the workspace tests pass. The exceptions are outbound_refill_journey, which needs sudo for a netns locally, and tip_accept::two_full_senders_both_enqueue, which failed once under workspace load and passed 8/8 alone and in a full rbitcoin-net suite run.

This changes the Core functional harness, so it needs the core-functional label. I cannot add labels on this repo.

🤖 Generated with Claude Code

[rearden-grok]

APPROVE

0 bugs, 0 suggestions, 0 nits.

[reardencode]

I hate this PR because we don't have a UTXO set, but agreed it's the right compatibility thing to do. Thanks!