fix: Use filename w/extension

[ns-rse]

Closes #1290
Closes #1287

Because of problems that arise with the old format of Bruker files where the filenames have numerical extensions there
was scope for data being over-written as TopoStats.filename uses the "stem", the part upto the file extension and so
file1.001, file1.002 and file1.003 all write output to the same (nested) directory file1. To solve this we now
include the file-extension in TopoStats.filename.

However, this presents a problem when processing .topostats files as you can't have a directory and a file with the
same name in the same location (manifested in the tests where we use minicircle_small.topostats). The solution is
therefore to write output files (the .topostats version of every file) to the directory
<output_dir>/processed/topostats/<filename_without_extension>.topostats.

In doing so we also modify io.save_topostats_file() and drop the filename parameter as it can be obtained directly
from the topostats_object.filename parameter passed in. We also introduce a step when saving files that don't already
carry the .topostats extension (i.e. .spm, .ibw, .006 etc.) to remove the last file extension from the filename
and replace with .topostats for the target that is to be written to.

This hopefully also addressed #1290 at the same time @tcatley if you are able to check with the
20240411_60ngTop2a_6ngpuc19_1mMATP_mgni.0_00024 file that uncovered the problem please that would be really useful and
appreciated.

---

Before submitting a Pull Request please check the following.

• Existing tests pass.
• Pre-commit checks pass.

[tobyallwood]

Looks good to me 👍

[tcatley]

This does run and output the .csv files. However as @tobyallwood pointed out in #1290, there is now two folders generated for each file - one with the file name with the .spm extension (containing the tracing/filters/grains) and one without (containing the csv files). Ideally this would only be one folder per image

[ns-rse]

Thanks @tcatley I've now fixed that, took longer than I expected (as is often the case!).

It stemmed from the disconnect between the code that writes all results out and the section that writes out per image statistics.

I do wonder though, as I had this problem when working on #1284, what the utility of duplicating output is. The all statistics .csv files at the top level of the output_dir contain the same results as the per-image, it just requires filtering. Maintaining such convenience functionality detracts from developer time to work on other tasks.

Basic data manipulation skills are invaluable and something all scientific researchers should have the in their skill set. An alternative if a single images output really is required is to just re-run TopoStats on just the single file.

[tobyallwood]

@ns-rse could it be possible the other set of files are left over from when we changed the names a few months back? Will take a look at the new code later am just out right now

[tcatley]

Agreed @ns-rse, I can't say I've ever actually used the per-image stats anyway - I would always just filter from the all_statistics.csv. If it is easier to just remove that output folder completely then maybe go with that

[ns-rse]

@tobyallwood : No there is a specific function for writing them on a per image basis (it uses the basename value from the results dataframes) so it was something that was deliberately done at some point.

@tcatley : Good to hear, thanks. I'll write up removal as an issue to be addressed.