Nice work on decoupling sweep power from the TX applet Tune Power, @jensenpat. The PGXL safety gate is a good addition — preventing an SWR sweep from driving RF into an amplifier in operate mode is the right call. A few observations:

Looks good

• Sweep power save/restore lifecycle — saving sweep power to AppSettings, applying it before RF, and restoring originalTunePower after finishSwrSweepAfterTuneStopped is clean and correct. The restore is positioned before the TGXL restore branch so it always runs.
• QSignalBlocker in setSwrSweepPowerWatts — prevents feedback loops when syncing the slider across panadapters. Good.
• const auto& tuner in advanceSwrSweep — correct, the sweep timer path is read-only there.
• Suppressing the empty SWR plot (SpectrumWidget.cpp) — removing the && !m_swrSweepRunning check is the right fix; no reason to enter the draw path with zero points.
• AppletPanel includes — QContextMenuEvent / QMenu additions look like they were already needed by existing code but missing; fine to include here.

Items worth a look

1. setTunePower optimistic update — The new code updates m_tunePower locally and emits stateChanged() before the command reaches the radio. This is fine for the sweep's immediate needs, but note that setRfPower() does not do this — the two setters now behave asymmetrically. If the radio responds with a status update carrying the same value, the status handler at line 46–48 will no-op (no double-emit), so there's no functional bug. But if the command fails silently (radio rejects the power level), the local model will be out of sync until the next status push. Worth a comment noting the intentional asymmetry, or consider reverting the model on command failure if that's feasible. Not blocking.

2. Tune power restore on abort — If the sweep is aborted before tx.setTunePower(sweepPowerWatts) is reached (e.g., TGXL bypass wait times out and the sweep never starts RF), finishSwrSweepAfterTuneStopped will still call setTunePower(originalTunePower). Since originalTunePower == tx.tunePower() at that point (the sweep power was never applied), the originalTunePower != sweepTunePower guard correctly prevents a no-op restore — nice. Just wanted to confirm I traced that right.

3. kSwrSweepPowerSettingKey is constexpr const char* — Minor: in C++20 you could use constexpr auto kSwrSweepPowerSettingKey = "SwrSweepPowerWatts"; or constexpr std::string_view. The current form is fine and consistent with the existing constants above it, so no action needed unless you want to tighten it.

4. Signal signature change — swrSweepStartRequested(int, int) is a breaking signal change. Since this is internal (no QML/plugin consumers), it should be fine, but if any out-of-tree code connects to this signal by name they'll get a runtime connect failure. Just flagging.

Overall this is solid and well-scoped. Thanks for the contribution!