Add support for empty value in path fieldtype

[yunzheng]

Normally an empty path would be normalized to a "." (dot) character.
This change allows you to initialize a path field with an empty string.
This is useful to represent a path that is empty.

Fixes DIS-2557

[yunzheng]

Currently flow.record.fieldtypes.path() (without any arguments) will also be considered an empty path. Not sure if this is preferred or to only consider an empty path if it's initialized with an empty string "".

[yunzheng]

No idea why Python 3.12 fails, will look into it later

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 83.72%. Comparing base (43a5656) to head (d5f8264).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #122      +/-   ##
==========================================
+ Coverage   83.67%   83.72%   +0.04%     
==========================================
  Files          34       34              
  Lines        3394     3403       +9     
==========================================
+ Hits         2840     2849       +9     
  Misses        554      554              

Flag	Coverage Δ
unittests	83.72% <100.00%> (+0.04%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[cecinestpasunepipe]

Your solution is somewhat similar to my original proposal, which was rejected because it failed the following test (which also fails here):

 # initialize with empty path
    p1 = path_cls(Path(""))
    assert p1 == ""
    assert p1._empty_path
    assert str(p1) == ""
    assert p1 != path_cls(".")

[yunzheng]

That is because you initialise a flow.record path with a stdlib Path(""). I think the original PR was rejected because we don't want to change the behaviour of pathlib.Path.

[cecinestpasunepipe]

According to my information this was part of the requirements. I see no other way to implement it.

[Schamper]

As long as this works with our own TargetPath, all should be well in the world. And even then, we generally just directly call windows_path(variable) anyway. Having side-effects is a big blocker, so if that's really the only way, then we should just document this behaviour and force usage through this way.

[yunzheng]

There was not much information regarding the requirements in the DIS ticket, other than that Record path fields do not support empty values as they would end up as the . dot value. This PR fixes that.

If windows_path and posix_path fields are intitialised using TargetPath variables, then we need to ensure that TargetPath also supports empty paths and then we can test if that works.