Sleep during sensor reading methods

[joshmcculloch]

Hi, I'm looking for justification for the sleep functions in the sensor reading methods.
There is a commit from May 2020 stating they were added to fix timing issues. But I wonder if they really needed to be added to the sensor reading methods. They significantly reduce the max sample rate on the raspberry pi, and I haven't had any issues since removing them:
516f2dc
I'd like to create a pull request unless I have mistaken their purpose.

[ladyada]

you could do a PR, and we may test/merge when we can!

[jposada202020]

Hello, according to my tests, there could be 3 situations that timing is important:

1. After, Reset (soft-reset) DEVICE_RESET in the PWR_MGMT_1 register
2. After Sleep
3. After reading a different memory bank. But, not sure about this.

Maybe the logic behind removing the sleep from the acc, gyro readings was that probable you could read the 6 values (7 with temp) in one shot, so time was less sensitive there.

Disclaimer: I have not tested the ICM20649.