add known issues in statistics guide

[loretet]

Added a warning to the "Statistics guide" section of the documentation to include two known issues, #2278 and #2378.

Both issues are not proper bugs and can be easily worked around, but I think the user should still be made aware of their existence (otherwise both issues can silently mess up the user analysis, more or less significantly)

[adperezm]

Thanks for putting this up!

Isn't it best to simply recommend to always use weighted averages for processing any of the statistics from Neko? Is there any case where that would fail?

I would say that when post-processing the data, it is important to inspect the results. However, I do not see the harm in giving this information.

On the other hand, I am still not sure that these things are actually read.

[vbaconnet]

Thanks for opening the PR!

I agree with Adalberto's comments.

The two "issues" you have raised only appear if you use an arithmetic mean for the final averaging. Therefore I would suggest that we:

• recommend the user use weighted averages exclusively, as is done in the provided post-processing script average_fields_in_time
• inform the user that a zero-filled stats file will always be dumped at (or slightly after) start_time so they know what to expect.
• Maybe refer to the documentation for the case file to remind them about the behavior of output_at_end (also not an issue if using weighted averaging).

[vbaconnet]

Hi, I pushed some stuff just to reformat a bit with the 80 column limit.

[vbaconnet]

To make it more general, the name "stats0" is specific to the csv format

Suggested change

	entry in the `stats0` file will be zero-filled (both for scalar and fluid
	sample will be zero-filled (both for scalar and fluid

[vbaconnet]

Thanks for making all the chnages! I added two comments but they are quite minor, up to you if you want to change or not