RF Scan/copy: keep every capture in a selectable list

[vnxdtzip]

Proposed Changes

Scan/copy currently shows one signal at a time. Every new capture overwrites the previous one on screen, and [NEXT] opens a single global menu that acts on whatever happened to be the last decode — so capturing a second remote means losing the first, and there is no way to go back to a signal you already caught.

This PR turns the capture screen into a list, following the same interaction model used by the file browsers elsewhere in Bruce:

• Captures accumulate in an on-screen list (10 slots). Each row shows the protocol and key (1. CAME 1A2B (12b)), the manufacturer and counter for KeeLoq (3. KeeLoq Nice cnt=42), or the pulse count for raw captures (2. RAW 148 pulses), with a * marking signals already written to storage.
• A compact status line keeps the frequency/range, the mode (Decode / RAW / RAW+CRC), an A when auto-save is on, and the n/10 counter.
• PREV/NEXT (or UP/DOWN) scroll the list, SEL opens a per-signal menu — Signal Info, Replay, Replay as RAW, Save Signal, Save as RAW, Delete — and ESC exits.
• The global settings (Range, Threshold, Mode, Filter, auto-save, Clear List, Main Menu) moved into a Scan Options entry, also reachable directly with SEL while the list is still empty.

Behaviour changes that come with it:

• Duplicate captures are dropped (same protocol/key/preset/bits). Holding a remote button no longer fills the list with copies of the same frame. KeeLoq is unaffected, since its key changes on every press. Undecoded RAW frames have no key and are always kept, because timing jitter makes them impossible to compare reliably.
• Replay/Save act on the selected entry instead of a single shared buffer, and keeloq_step(1) now advances the counter of that specific entry.
• Screen-only actions (Info, Delete, Clear List, closing a menu) no longer pay the deinitRfModule() + 1 s settle that every menu action used to trigger; they just re-arm the receiver. Actions that transmit, reconfigure the radio or touch the shared SPI bus keep the full re-init.
• The Signal Info view waits for a genuinely new key press. Some input drivers — the Cardputer ADV's TCA8418 among them — raise AnyKeyPress on key release as well, so the leftover release event from the menu selection dismissed the screen the instant it was drawn.

The list is capped at 10 entries because a RAW slot holds a duration string of up to ~3 KB (the RMT session captures 256 symbols), which keeps the feature usable on the boards without PSRAM.

Types of Changes

Enhancement, plus one bugfix (Signal Info dismissed immediately on TCA8418 keyboards). No breaking changes to the .sub format, the RF decoders or the serial commands — display_info(), rfReceiveSignal() and rf_scan() are untouched and still serve the CLI/headless paths.

Verification

1. RF -> Scan/copy, press a 433 MHz remote a few times: one entry appears per distinct signal, repeats are ignored.
2. Press a different remote: it lands as a second entry, the first one is still there.
3. Scroll with PREV/NEXT, press SEL on an entry -> per-signal menu.
4. Replay: the selected signal is transmitted (not the last one captured).
5. Save: the entry gets a *; the .sub file is written under /BruceRF as before.
6. Delete: the entry disappears and the list renumbers.
7. Signal Info: the details stay on screen until a key is pressed.
8. With the list empty, SEL opens Scan Options directly.

Tested by flashing m5stack-cardputer and elecrow-advance-35-s3; both compile clean and the screen behaves as described. Also exercised on a touch-only board, where the 3x3 touch heatmap maps to PREV/NEXT/SEL/ESC.

Testing

Not covered by automated tests — the change is UI/interaction code driven by the display and the input task, which the repo does not currently have a harness for. The signal decoding and .sub writing paths it calls into are unchanged.

Linked Issues

None.

User-Facing Change

RF Scan/copy now keeps every captured signal in a scrollable list instead of showing only the last one. PREV/NEXT scroll, SEL opens a per-signal menu (Info, Replay, Replay as RAW, Save, Save as RAW, Delete); the previous global options moved to a "Scan Options" entry. Duplicate captures of the same frame are ignored.

Further Comments

Two things worth a maintainer's opinion:

• Button mapping. [NEXT] no longer opens the options menu — it scrolls the list, and SEL opens the menu. This is the convention already used by the file browsers, but it is a change in muscle memory for Scan/copy users. On HAS_3_BUTTONS boards, where PREV and ESC share a button, the same guard loopOptions() uses is applied so a PREV tap does not quit the scan.
• List size. 10 is a compromise for the boards without PSRAM. If that is too conservative, it is a single #define _MAX_CAPTURED in rf_scan.h.

The commit also replaces the setup()/loop() mutual recursion (and the options-menu recursion) with plain loops — the old code relied on tail-call optimisation to avoid growing the stack on every menu open.

🤖 Generated with Claude Code

[vnxdtzip]