perf(FormulaDict): sort lazily instead of on every assignment

[rkingsbury]

Summary

FormulaDict (backing Solution.components) re-sorted its entire contents on every __setitem__. A burst of writes — populating a Solution, or the volume setter's self.components[s] *= factor loop — was therefore O(n² log n).

This defers the sort: a mutation just marks the dict dirty, and the sort happens lazily at the iteration barrier (__iter__, plus __repr__ since UserDict.__repr__ reads self.data directly). Callers still see the same descending-by-amount order, because keys()/values()/items()/list()/dict()/for all funnel through __iter__. Deleting a key preserves relative order, so it leaves the flag untouched.

No behavior change — iteration and repr order are identical to the previous eager sorting; only the timing of the sort moves.

Benchmark — building all 23 presets from test_from_preset

Metric	Before	After	Change
FormulaDict re-sorts / full pass	5843	94	~62x fewer
Isolated build, median (all presets)	38.06 s	29.67 s	~22% faster
pytest test_from_preset wall time	87.00 s	71.51 s	~18% faster

Testing

• test_from_preset — 23/23 pass.
• test_component_subsets — directly asserts the descending-amount ordering contract of cations/anions/neutrals; passes.
• test_formula_dict, test_charge_balance, test_get_amount, test_components_by_element(_nested), test_arithmetic_and_copy, test_from_dict_complex, and the serialization tests — all pass.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.81%. Comparing base (f8d582b) to head (4f14b17).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #452      +/-   ##
==========================================
+ Coverage   87.75%   87.81%   +0.06%     
==========================================
  Files          14       14              
  Lines        1935     1945      +10     
  Branches      337      338       +1     
==========================================
+ Hits         1698     1708      +10     
  Misses        188      188              
  Partials       49       49              

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.