Noticeable CPU load on low-power Pi

[floe]

Describe the bug
I'm running Theengs Gateway on a Raspberry Pi Zero, and getting pretty consistent CPU load of ~ 35%.

To Reproduce
Steps to reproduce the behavior:

1. Install TG in a venv on a Raspberry Pi Zero.
2. Run it, observe CPU usage.

Expected behavior
Way less CPU load (IMHO). From looking at the log, it's blasting out somewhere around 100 MQTT messages per second, and I only have one single plant sensor at the moment.

Environment

# Theengs Gateway Diagnostics

## Package Versions

| Name               | Value  |
|--------------------|--------|
| Theengs Gateway    | 1.4.0  |
| Theengs Decoder    | 1.7.2  |
| Bleak              | 0.21.1 |
| Bluetooth Clocks   | 0.2.0  |
| Bluetooth Numbers  | 1.1.1  |
| Paho MQTT          | 2.0.0  |
| Bluetooth Adapters | 0.18.0 |

## Python

| Name           | Value                          |
|----------------|--------------------------------|
| Version        | 3.11.2                         |
| Implementation | CPython                        |
| Compiler       | GCC 12.2.0                     |
| Executable     | /home/floe/theengs/bin/python3 |

## Operating System

| Name         | Value                                    |
|--------------|------------------------------------------|
| System       | Linux                                    |
| Release      | 6.6.20+rpt-rpi-v6                        |
| Version      | #1 Raspbian 1:6.6.20-1+rpt1 (2024-03-07) |
| Machine type | armv6l                                   |
| Distribution | Raspbian GNU/Linux 12 (bookworm)         |

## Configuration

File: /home/floe/theengsgw.conf

{
    "adapter": "",
    "bindkeys": {},
    "ble": 1,
    "ble_scan_time": 5,
    "ble_time_between_scans": 5,
    "discovery": 1,
    "discovery_device_name": "TheengsGateway",
    "discovery_filter": [
        "IBEACON"
    ],
    "discovery_topic": "homeassistant",
    "enable_tls": 0,
    "enable_websocket": 0,
    "general_presence": 0,
    "hass_discovery": 1,
    "host": "jarvis.lan",
    "identities": {},
    "log_level": "INFO",
    "lwt_topic": "home/TheengsGateway/LWT",
    "pass": "***",
    "port": 1883,
    "presence": 0,
    "presence_topic": "home/TheengsGateway/presence",
    "publish_advdata": 0,
    "publish_all": 1,
    "publish_topic": "home/TheengsGateway/BTtoMQTT",
    "scanning_mode": "active",
    "subscribe_topic": "home/+/BTtoMQTT/undecoded",
    "time_format": 0,
    "time_sync": [],
    "tracker_timeout": 120,
    "user": "***"
}

## Bluetooth adapters

Default adapter: hci0

### hci0

| Name         | Value                   |
|--------------|-------------------------|
| address      | B8:27:EB:XX:XX:XX       |
| sw_version   | fjordcam                |
| hw_version   | usb:v1D6Bp0246d0542     |
| passive_scan | False                   |
| manufacturer | Raspberry Pi Foundation |
| product      | None                    |
| vendor_id    | None                    |
| product_id   | None                    |

Additional context
I assume that there is some config file or commandline switch I could use to slow it down - are the -sd and -tb options related to that?

[floe]

Additional note: I tried with -pa 0 -tb 10 -sd 2 and that has definitely cut down the number of MQTT messages being sent, but doesn't seem to have had any noticeable impact on CPU load.