Add a (optional) way to measure CPU load on Cortex M executor

[spcan]

This PR adds an opt-in way to create an Executor with very basic CPU load measurements. An example for the STM32F7 MCU is included.

Pros:

• Builtin method that measures CPU load of a poll cycle
• Each executor has its own independent measurement, no synchronization needed
• It's opt-in, no changes to the executor if the feature is not used
• No additional dependencies

Cons:

• Adds a small delay before the executor starts polling and before the executor goes to sleep
• The data provided is very simple, the user must implement their own logic to use the data properly
• Not compatible with #[embassy_executor::main] macro