v0.10.2

Added

• The published CLI is installed and held to its documented contract
  (#326). install.yml verified cargo add termlens and never
  cargo install termlens-cli, so #310 — termlens inspect --version
  exiting 2 — shipped in 0.10.0, shipped again in 0.10.1, and was found by a
  user of the published binary rather than by anything here.

  A cli job installs termlens-cli from crates.io on Linux and macOS —
  every run, never from a cache — and asserts every exit code
  termlens --help documents: --version in all four positions returning
  the same string, inspect printing a screen with its header and trailer,
  render in all four formats, and diff returning 0 for the same
  picture, 1 for a different one and 2 for input it cannot read.

  The assertions live in .github/scripts/check-cli-contract.sh and take
  the binary to exercise, so the same script runs against a path build. ci.yml
  does exactly that on every pull request: the published check is the net,
  the tree check is the fast feedback. Re-introducing #310 locally was
  observed to fail it.

Fixed

• unsupported() no longer names the four SGR parameters the attribute
  shadow implements (#320). 5/25 (blink), 9/29 (strikethrough),
  and 6/8/28 with them are exactly what emu/shadow.rs exists to
  recover: vt100 drops them, the shadow parser puts the attribute on the
  cell, and the tracker then reported the backend's gap under an accessor
  documented as naming the emulator's.

  The consequence was the one the accessor was added to prevent, inverted: a
  screen said Style::blink was true and unsupported() said ^[[5m had
  been dropped, at the same instant — so a reader checking the list before
  trusting a blink or masked-password assertion concluded a correct
  assertion was unreliable, and
  assert!(screen.unsupported().is_empty()) could never pass for an
  application that uses either attribute.

  Found by termlens-demo driving a real ratatui application against the
  published 0.10.1, which is what the testing tier is for.

  What still gets named. An SGR with one parameter nobody models keeps
  the whole sequence: ^[[59m (underline colour) alone, and ^[[5;59m,
  where blink is recovered and 59 is not. Dropping a mixed sequence
  because part of it is implemented would hide a real gap.

  The residue, measured rather than assumed. ^[[1;5;31m is still
  named although bold, red and blink all reach the cell — 1 and 31
  are vt100's to implement, and this tracker knows what the shadow
  recovers, not what the backend does. Narrowing that needs vt100's own SGR
  surface enumerated, which is more than a patch should claim; a test
  records the behaviour so it is a documented edge rather than a surprise.

• termlens <subcommand> --version is no longer an unknown option.
  The flag was handled only in the top-level dispatch, so termlens inspect --version exited 2 while -h/--help worked in that position. All three
  subcommands now print the same version string as the top-level command. (#310)