Telemetry and the LUA script works fine showing "other devices" and loading the RX settings even under heavy telemetry load with some expected limitations:

All Full-res packet rates in all available Switch Modes and all telemetry ratios,
all packet rates in Hybrid Mode and all telemetry ratios,
all packet rates in Wide Mode and telemetry ratios 1:2, 1:4:

• RX settings loading times are comparable to pre-3.5.5 firmware versions

All packet rates in Wide Mode and telemetry ratios 1:8 or slower:

• RX settings loading times are as expected about factor 3 (50Hz about factor 4) slower than compared to pre-3.5.5 firmware version

This is working flipping great, amazing refactor with support for a longer queue to support bursts. Most of these are minor things, but we do need to address the webui not loading for me on 8285 RXes (2 resources load and then the web server never responds to rest) but we can maybe tackle that in a separate PR. Here's what I see, and I can push most of these changes.

• Bug introduced in lua.cpp pushResponseChunk(), should be else nextStatusChunk++;
• dummy[CRSF_MAX_PACKET_LEN] -> replace messagePayloads.popBytes(dummy, sz); with skip()
• Remove crsf_telemetry_package_t and member
• Remove receivedPackages
• Remove twoslotLastQueueIndex
• Remove currentPayloadIndex
• Comment for ACTION_NEXT suggest "continue searching queue for other messages"
• settingsCount should be member, and probably named prioritizedCount?
• bit(7) should be a define and use e.g. IS_DEL(sz) SET_DEL(sz) SIZE(sz) etc
• FIFO<TELEMETRY_FIFO_SIZE> everywhere, typedef it
• Should comparator_t definition have const as well?
• messagePayloads[] data is accessed without lock
• uint16_t messagePosition = 0 -> rename overwritePosition
• if (whatToDo == ACTION_IGNORE || whatToDo == ACTION_APPEND || whatToDo == ACTION_OVERWRITE) is clearer if it is if (whatToDo != ACTION_NEXT)
• Add comparitor for CRSF_FRAMETYPE_PARAMETER_SETTINGS_ENTRY to check for duplicates (OVERWRITE)
• Should AppendTelemetryPackage also check currentPayload? Considering duplicates of PARAMINFO being requested while that value is currently being transmitted. Would require changing signature to use some sort of accessor helper class
• MicroOptimization: If the first item in the queue is prio, use pop instead of marking deleted. This only triggers at high packet rates so probably doesn't matter?
• Test for two different deviceinfo
• Test function that doesn't require popping and requeing everything

int Telemetry::UpdatedPayloadCount()
{
    int retVal = 0;
    unsigned pos = 0;
    while (pos < messagePayloads.size())
    {
        uint8_t sz = messagePayloads[pos];
        if (!(sz & bit(7)))
        {
            retVal++;
        }
        pos += (sz & 0x7F) + 1;
    }
    return retVal;
}

• Add uint8_t payload[0] to crsf_header_t and use payload[0] instead of ((uint8_t*)newMessage)[CRSF_TELEMETRY_TYPE_INDEX + 1]
• Move CRSF_FRAMETYPE_OPENTX_SYNC to be in a new enum, subtype?
  • CRSF_FRAMETYPE_RADIO_ID rename to CRSF_FRAMETYPE_HANDSET? (spec lists as Remote Related Frames)
  • CRSF_FRAMETYPE_OPENTX_SYNC go to subcommand rename to CRSF_HANDSET_SUBCMD_TIMING
• Can we reuse deleted slots to reduce buffer use? Push can also defragment as it goes, merging an adjacent deleted block into one larger deleted block
• CRSF_AP_CUSTOM_TELEM_STATUS_TEXT is actually a log so both our old and new mechanism isn't great for servicing this message, but that's outside the scope of this PR