Update Measuring_Resonances.md for some NumPY version issues

[JamesH1978]

It has been noted over the last six to eight months that some versions of Numpy have issues with the klipper python environment on some machines. This PR introduces a fixed version that is known to work and a small test for people to do to make sure there are no output issues from the get go. These have been pulled from the pinned posts in the discord, from a time when 1.26 was causing issue, and now it seems v2 is also having some issues, hence the change.

Thanks James

[dmbutyugin]

In general, I have no concerns about this change, as we are not using any special features from more recent releases. However, should we really force a specific version (1.25.2). Would it make sense to specify either ~=1.25.0 or <1.26 ? TBH, I do not really mind it either way.

[JamesH1978]

I believe 1.26 was the first version to show issues, and 1.25.2 was the last known not to basically

[dmbutyugin]

Sorry, I was probably unclear with my comment. What I meant was, instead of fixing a very specific version, wouldn't it be better to let pip choose the best available version (for a given version of installation) prior to 1.26?

[JamesH1978]

Ah sorry, i reread and got your meaning now! yes i could do less than 1.26

[KevinOConnor]

Okay, thanks. I'll give a few days to see if there are other comments, and otherwise look to commit.

-Kevin

[Sineos]

Thanks @JamesH1978. While you are already working on this document, please check https://klipper.discourse.group/t/documentation-update-for-input-shaper-page/19149

[JamesH1978]

@Sineos yes we have seen a few people wonder why the docs don't say the right thing, when they don't follow them correctly ;) to be fair that doc needs a complete rewrite and maybe splitting out into a separate doc for adxl, mpu and lis, as it has become a tome.

[Sineos]

I wholeheartedly agree @JamesH1978 Guess it is the same as with software: Features are added, bugs are fixed and at a certain point you need to start refactoring as it has become hard to maintain. Still, I think, Klipper has an outstanding documentation overall, given the complexity of the product.

In this case, it would be a small bug fix by exchanging /tmp/resonances_x_*.csv with /tmp/calibration_data_x_*.csv

Anyway, thank you for taking a little care of the documentation 👍

[JamesH1978]

Its not a bug though :) the instructions that give that line are linked to the TEST_RESONANCES command, not SHAPER_CALIBRATE, where the former does output resonances_x_*.csv, unlike the latter which outputs calibration_data_x_*.csv. There is no part of this doc that references a png conversion for SHAPER_CALIBRATE. It could however be added that the png's could also be created with a switched up command referencing the correct file

[Sineos]

You are right. Everywhere the preceding command is indeed TEST_RESONANCES when it is about using the data files to generate a graph afterwards.

[KevinOConnor]

Thanks.

-Kevin