feat: Allow subtraction of time dtype columns

[ptiza]

Closes #19720

Since the documentation specifies the Time dtype as

The underlying representation of this type is a 64-bit signed integer. The integer indicates the number of nanoseconds since midnight.

we can specify the result of a substraction as a duration with nanosecond time unit.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.59%. Comparing base (bcca075) to head (32a2475).
Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #20300      +/-   ##
==========================================
- Coverage   79.60%   79.59%   -0.02%     
==========================================
  Files        1565     1566       +1     
  Lines      218396   218322      -74     
  Branches     2477     2465      -12     
==========================================
- Hits       173860   173778      -82     
- Misses      43970    43975       +5     
- Partials      566      569       +3     

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

[mcrumiller]

What about the inverse? If we have t2 - t1 = d, how about t1 + d = t2, i.e. time + duration?

[ptiza]

What about the inverse? If we have t2 - t1 = d, how about t1 + d = t2, i.e. time + duration?

I considered implementing it within this PR but thought it's ultimately another feature.

It also needs to be discussed which behavior we want if d > 24h. Should we even allow it?
If d is the result of a substraction of two Time objects, then the resulting duration is always within 24h.

[mcrumiller]

I recall a while back during some AnyValue conversion that @stinodego came up with a set of rules regarding what temporal ops are allowed.

t2 - t1 will always be a well-defined duration value <= 24hr, but t + d may be out-out-bounds and so perhaps is not as well-defined.