Improve precisedelta to better support rounding and float values.

[nuztalgia]

Fixes #14.
Fixes #20.

Changes proposed in this pull request

• Implement better support for rounding and float values in precisedelta.
• Add the test cases listed below, to clarify/enforce desired behavior.
• Improve grammar and consistency in some internal docstrings/comments/annotations.

Test cases added to test_precisedelta_custom_format

>>> precisedelta(dt.timedelta(days=31), "days", format="%d")
BEFORE:  "1 month and 0 days"
AFTER:   "1 month"

>>> precisedelta(dt.timedelta(days=31.01), "days", format="%d")
BEFORE:  "1 month and 0 days"
AFTER:   "1 month and 1 day"

>>> precisedelta(dt.timedelta(days=31.99), "days", format="%d")
BEFORE:  "1 month and 1 days"
AFTER:   "1 month and 1 day"

>>> precisedelta(dt.timedelta(days=32), "days", format="%d")
BEFORE:  "1 month and 1 days"
AFTER:   "1 month and 2 days"

>>> precisedelta(dt.timedelta(days=62), "days", format="%d")
BEFORE:  "2 months and 1 day"
AFTER:   "2 months and 1 day"  # no change

>>> precisedelta(dt.timedelta(days=92), "days", format="%d")
BEFORE:  "3 months and 0 days"
AFTER:   "3 months"

>>> precisedelta(0.01, "seconds", format="%0.3f")
BEFORE:  "0 seconds"
AFTER:   "0.010 seconds"

>>> precisedelta(31, "minutes", format="%d")
BEFORE:  "0 minutes"
AFTER:   "1 minute"

>>> precisedelta(60 + 29.99, "minutes", format="%d")
BEFORE:  "1 minutes"
AFTER:   "1 minute"

>>> precisedelta(60 + 30, "minutes", format="%d")
BEFORE:  "1 minutes"
AFTER:   "2 minutes"

>>> precisedelta(60 * 60 + 30.99, "minutes", format="%.0f")
BEFORE:  "1 hour and 0 minutes"
AFTER:   "1 hour"

>>> precisedelta(60 * 60 + 31, "minutes", format="%.0f")
BEFORE:  "1 hour and 1 minutes"
AFTER:   "1 hour and 1 minute"

>>> precisedelta(ONE_DAY - MILLISECONDS_1_337, "seconds", format="%.1f")
BEFORE:  "23 hours, 59 minutes and 58 seconds"
AFTER:   "23 hours, 59 minutes and 58.7 seconds"

>>> precisedelta(ONE_DAY - ONE_MILLISECOND, "seconds", format="%.4f")
BEFORE:  "23 hours, 59 minutes and 59 seconds"
AFTER:   "23 hours, 59 minutes and 59.9990 seconds"

Test cases added to test_precisedelta_multiple_units

>>> precisedelta(dt.timedelta(days=31), "seconds")
BEFORE:  "1 month and 0 days"
AFTER:   "1 month and 12 hours"

>>> precisedelta(dt.timedelta(days=32), "seconds")
BEFORE:  "1 month and 1 days"
AFTER:   "1 month, 1 day and 12 hours"

>>> precisedelta(dt.timedelta(days=62), "seconds")
BEFORE:  "2 months and 1 day"
AFTER:   "2 months and 1 day"  # no change

>>> precisedelta(dt.timedelta(days=92), "seconds")
BEFORE:  "3 months and 0 days"
AFTER:   "3 months and 12 hours"

>>> precisedelta(dt.timedelta(days=31), "days")
BEFORE:  "1 month and 0.50 days"
AFTER:   "1 month and 0.50 days"  # no change

>>> precisedelta(dt.timedelta(days=32), "days")
BEFORE:  "1 month and 1.50 days"
AFTER:   "1 month and 1.50 days"  # no change

>>> precisedelta(dt.timedelta(days=62), "days")
BEFORE:  "2 months and 1 day"
AFTER:   "2 months and 1 day"  # no change

>>> precisedelta(dt.timedelta(days=92), "days")
BEFORE:  "3 months and 0.50 days"
AFTER:   "3 months and 0.50 days"  # no change

[codecov-commenter]

Codecov Report

Merging #39 (e2f2dad) into main (e705e43) will decrease coverage by 2.58%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main      #39      +/-   ##
==========================================
- Coverage   99.13%   96.54%   -2.59%     
==========================================
  Files           9        9              
  Lines         690      694       +4     
==========================================
- Hits          684      670      -14     
- Misses          6       24      +18     

Flag	Coverage Δ
macos-latest	96.54% <100.00%> (-0.85%)	⬇️
ubuntu-latest	?
windows-latest	?

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

Impacted Files	Coverage Δ
src/humanize/time.py	99.54% <100.00%> (+<0.01%)	⬆️
tests/test_time.py	100.00% <100.00%> (ø)
src/humanize/__init__.py	80.00% <0.00%> (-20.00%)	⬇️
tests/test_i18n.py	87.50% <0.00%> (-12.50%)	⬇️
src/humanize/i18n.py	90.56% <0.00%> (-7.55%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e705e43...e2f2dad. Read the comment docs.

[nuztalgia]

(Force-pushed to clean up the commit history, since it was getting a bit long. 😅 Let me know if you'd like me to squash it down even more!)

This PR now includes one of the commits from #19, to fix the root cause of #14. Thank you, @eldipa! 💜

I've also updated the description (#39 (comment)) to reflect all of the test cases that this PR adds and fixes.