Refactor utility/model hot paths into leaner drop-in implementations

[Copilot]

This Pull Request tightens core utility/model paths with behavior-preserving refactors focused on lower allocation, less duplication, and simpler control flow. It keeps the public surface and expected runtime behavior intact while reducing maintenance and execution overhead.

• Scope

  • Narrow refactor only in shared helpers used across routing, deep-link parsing, and analytics unit adaptation.

• Key changes

  • Consolidated duplicated DB URL resolution into one internal helper (configured_exercise_db_url) used by both DB and image base URL accessors.
  • Reused a single CSV ID parser for deep-link session-start flows instead of duplicated split/filter/map blocks.
  • Optimized query percent-decoding to avoid allocation when + is absent (borrowed fast path via Cow).
  • Switched signed time formatting to unsigned_abs() for safer/intention-revealing negative handling.
  • Simplified exercise-type tag selection with a compact match on category/force semantics.
  • Kept explicit empty-input handling for analytics average computation to preserve semantic clarity.

• Example

fn percent_decode(s: &str) -> String {
    use std::borrow::Cow;
    let s = if s.contains('+') {
        Cow::Owned(s.replace('+', "%20"))
    } else {
        Cow::Borrowed(s)
    };
    percent_encoding::percent_decode_str(&s)
        .decode_utf8_lossy()
        .into_owned()
}

Engineering Principles

• PR only contains changes strictly related to the requested feature or fix,
  scope is focused (no unrelated dependency updates or formatting)
• This code totally respects README’s Engineering Principles

CI/CD Readiness

• Branch follows Conventional Branch: feat/…, fix/…, refactor/…, …
• Code is formatted with dx fmt; cargo fmt
• All checks pass, nix flake checks succeeds without warnings
  • Code compiles, dx build with necessary platform flags succeeds
  • cargo clippy -- -D warnings -W clippy::all -W clippy::pedantic
    produces zero warnings
  • All unit tests pass without warnings
    cargo llvm-cov nextest --ignore-filename-regex '(src/components/|\.cargo/registry/|nix/store)'
  • End-to-end tests pass maestro test --headless maestro/web
maestro test --headless maestro/android

[github-actions]

📊 Coverage Report

Lines: 3756/4996 (75.18014411529224%)

⏱️ Tests: 255 tests in 0.607s

Filename	Function Coverage	Line Coverage	Region Coverage	Branch Coverage
main.rs	22.00% (11/50)	58.94% (155/263)	60.56% (218/360)	- (0/0)
models/analytics.rs	0.00% (0/6)	0.00% (0/33)	0.00% (0/45)	- (0/0)
models/enums.rs	100.00% (28/28)	100.00% (147/147)	100.00% (337/337)	- (0/0)
models/exercise.rs	92.00% (46/50)	92.10% (548/595)	91.01% (749/823)	- (0/0)
models/log.rs	100.00% (12/12)	100.00% (118/118)	100.00% (144/144)	- (0/0)
models/mod.rs	100.00% (11/11)	100.00% (67/67)	100.00% (97/97)	- (0/0)
models/session.rs	72.22% (13/18)	84.36% (151/179)	83.33% (210/252)	- (0/0)
models/units.rs	100.00% (28/28)	100.00% (167/167)	98.88% (353/357)	- (0/0)
services/app_state.rs	1.89% (1/53)	2.54% (11/433)	2.32% (15/646)	- (0/0)
services/exercise_db.rs	88.81% (127/143)	90.30% (1210/1340)	89.61% (1924/2147)	- (0/0)
services/exercise_loader.rs	0.00% (0/14)	0.00% (0/100)	0.00% (0/135)	- (0/0)
services/native_queue.rs	0.00% (0/15)	0.00% (0/145)	0.00% (0/197)	- (0/0)
services/notifications.rs	0.00% (0/3)	0.00% (0/9)	0.00% (0/10)	- (0/0)
services/service_worker.rs	100.00% (2/2)	100.00% (6/6)	100.00% (6/6)	- (0/0)
services/storage.rs	63.77% (88/138)	80.97% (766/946)	83.53% (1187/1421)	- (0/0)
services/wake_lock.rs	100.00% (2/2)	100.00% (5/5)	100.00% (5/5)	- (0/0)
utils.rs	91.36% (74/81)	91.42% (405/443)	91.78% (648/706)	- (0/0)
Totals	67.74% (443/654)	75.18% (3756/4996)	76.65% (5893/7688)	- (0/0)