fix: correct SecondsTimedeltaProvider for negative fractional seconds

[binggao1230]

Bug

SecondsTimedeltaProvider returns the wrong timedelta when loading a
negative fractional number of seconds.

from datetime import timedelta
from adaptix import Retort

retort = Retort()
print(retort.load(-0.5, timedelta))   # timedelta(0, 0, 500000)  — should be -0.5 s
print(retort.load(-1.5, timedelta))   # timedelta(-1, 86399, 500000)  — off by 1 s

Root cause

The loader uses:

timedelta(seconds=int(data), microseconds=int(data % 1 * 10**6))

int() truncates toward zero, so for data = -0.5:

• int(-0.5) → 0 (should be -1)
• -0.5 % 1 → 0.5 (Python modulo is always non-negative)
• result: timedelta(seconds=0, microseconds=500000) → +0.5 s ❌

Fix

Use math.floor() for the seconds component, which gives floor-division
semantics (always rounds toward negative infinity):

sec = math.floor(data)
timedelta(seconds=int(sec), microseconds=round((data - sec) * 10**6))

round() is used for the microseconds to avoid off-by-one float-precision
errors (e.g. -2.7 % 1 * 1e6 → 299999.99...). Decimal inputs are
handled correctly too since math.floor(Decimal(...)) returns an int.

Tests

Added four assertions for negative fractional float and Decimal inputs
to test_seconds_timedelta_provider. All 6 parametrized variants pass.

---

This pull request was prepared with the assistance of AI, under my direction and review.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
src/adaptix/_internal/morphing
concrete_provider.py					227
Project Total

This report was generated by python-coverage-comment-action

[zhPavel]

Thank you for your contribution!