AlgoScan

Adds runtime-argument calls for getbit/setbit/getbyte/setbyte with constant indices large enough to be statically impossible, restoring 100% line coverage after the generic-index branches were added.

Mirrors the recent getbit change: match a constant-bytes shape first (tight bound against the constant length), then fall through to a generic case that catches cases where the index is known constant but the first arg is runtime — index >= 8 * MAX_BYTES_LENGTH (or MAX_BYTES_LENGTH for byte ops) is statically impossible regardless of the runtime value.

Drop specific constant values from the reason strings so they describe the class of failure rather than the specific bad inputs, e.g.

  +: 18446744073709551615 + 1 overflows uint64; will fail at runtime if reached

becomes

  uint64 addition of constants overflows; will fail at runtime if reached

Specific values were useful while the validator was being bootstrapped but add noise for end users who can see the literal values at the source location anyway. Also drops the leading `{op.code}:` prefix since the op name is now folded into the reason text.

Same treatment as extract/extract3: each op has a fixed arg shape, so matching them separately lets the pattern encode the shape. Removes `_substring_start_end` and `_replace_start_and_bytes` helpers and their `*rest` unpacking plumbing.

Revealed a missing OP test case, so added coverage.

Each op has a fixed arg shape so matching them separately means the pattern itself encodes the shape — no `*rest` unpacking, no post-hoc `is not None` guards, and the `_extract_start_length` helper becomes dead and is removed.

Exercises every warning-emitting branch in statically_failing_op.py (both the top-level `_check` match arms and the three helpers' immediate-form / stack-form / None-fallthrough paths). Hits 100% line coverage of the validator via a single test case:

  coverage run -m pytest tests/test_compile.py -k statically_failing_ops -o addopts=""
  coverage report -m --include="src/puya/ir/validation/statically_failing_op.py"

Overlaps deliberately with a handful of the existing regression_tests cases since this is targeted at coverage completeness rather than regression pinning.