[vyncint]

A terminal that speaks kitty or sixel graphics moves the cursor past an image it places. termlens left the cursor where the image began, so text an application wrote after an image landed on top of it. docs/DESIGN.md also said that neither protocol moves the cursor.

The emulator now follows the terminal the test declared with graphics() and cell_size():

• kitty moves the cursor onto the placement's last row, one column past its right edge. It wraps past the screen's edge and scrolls past the bottom margin, unless C=1, U=1 or P= says otherwise. The size in cells comes from c=/r=. Without them it comes from kitty's own update_dest_rect arithmetic over the source rectangle and the cell size. a=p looks up its image's size by id in the payload log.
• sixel moves the cursor down to the image's last text row and keeps the column, as xterm does with sixel scrolling on. The height is the larger of the raster height and the painted height, measured before a payload past the capture bound is dropped.
• Graphics::None, the default, still moves nothing, as a terminal without the protocol ignores the escape.

The facts the move needs (a crate-internal Placement) travel beside the payload, not inside it, so the JSON shape is unchanged.

A kitty transmission that names no action is now a=t (transmit only), which is the protocol's default. It used to be reported as placed, which would also have moved the cursor for images kitty never shows.

LIMITATIONS.md says what is not modelled: sixel display mode ?80, xterm's extra scroll for a trailing -, ?8452, and a PNG placed without c=/r=.

Tests

• Unit tests in emu/vt100.rs:
  • kitty last row / one column past
  • wrap and scroll
  • pixel-sized placement via the cell size
  • sixel last text row with the column kept
• Unit tests in graphics.rs:
  • placement facts recorded
  • kitty_cells against kitty's arithmetic
  • sixel painted height
• PTY test a_declared_kitty_terminal_moves_the_cursor_past_a_placement in tests/graphics.rs. It adds a PTY test, so stress is needed.

Closes #588

Signed-off-by: Vyncint Ng 115854244+vyncint@users.noreply.github.com