ASERT DAA and timestamp hardening (activation: 2026-04-20)

[NyanCatTW1]

Goal

To ensure the chain recovers rapidly after major hashrate fluctuation, while keeping the chain safe from selfish mining attempts.
ASERT combined with hardened timestamp is safe from all except for 51% attacks, in which case we have better things to worry about than selfish mining

Summary

• ASERT difficulty adjustment: Port BCH's aserti3-2d algorithm, adapted for
  FACTOR's nonlinear (integer) nBits instead of BCH's 256-bit target. Uses a
  precomputed log2 lookup table with Q32.32 fixed-point arithmetic to compute
  2^(exponent) over nBits space. Half-life is 2 days on mainnet, 1 hour on
  testnet/regtest. Activates on mainnet at 2026-04-20 10:09:00 UTC.
• Timestamp hardening: Enforce monotonic block timestamps, remove peer-time
  adjustment (GetTimeOffset always returns 0), and reduce MAX_FUTURE_BLOCK_TIME
  to 4 seconds. ASERT uses block time directly rather than MTP for activation.

Details

Core implementation:

• CalculateFACTORASERT() - pure function, no chain-state deps (pow.cpp)
• asert_table.h - 512-entry log2(compute()) lookup table generated by
  scripts/gen_compute_table.py
• GetNextWorkRequired() routes to ASERT after activation

Timestamp changes:

• timedata.cpp ignores median peer offset
• validation.cpp enforces nTime > prev.nTime (monotonic)
• MAX_FUTURE_BLOCK_TIME reduced from 2h to 4s

Testing:

• asert_tests.cpp - unit tests for CalculateFACTORASERT including boundary,
  clamping, degenerate cases, and mainnet anchor
• pow_tests.cpp - updated for ASERT integration and monotonic timestamps
• gen_asert_test_vectors.cpp - deterministic test vector generator
• validate_nbits_aserti3_2d.py - independent Python validation of the
  fixed-point math

Smoke tests performed

• IBD still works
• Tested on testnet with 15s block time / 180s half life. Log from test is available here: testnet.txt