feat(core): broaden .understandignore starter — Swift XCTest/Quick patterns

[thejesh23]

Problem

.understandignore starter has no Swift-specific suggestions. Impact splits by project shape:

• SPM libraries (apple/swift, Alamofire, Vapor) already funnel all tests into Tests/ — caught by the case-insensitive tests dir rule. Little to gain.
• Xcode consumer apps (stripe-ios, kickstarter/ios-oss, duckduckgo/iOS, firefox-ios) use <AppName>Tests/ folders that no leading-dot suffix rule reaches, plus a heavy tail of inline *Tests.swift files scattered across production modules. Large gain — measurement showed up to 34% reduction on stripe-ios.

Proposal

New Swift group under TEST_PATTERN_GROUPS:

• **/*Tests.swift — dominant XCTest / Swift Testing convention
• **/*Test.swift — singular variant seen in older codebases
• **/*Spec.swift — Quick / Nimble BDD framework

All commented-out, opt-in. Same model as #76 / #479 / #578 / #581 / #596.

Deliberately not adding dir rules. Measurement across 20 Swift repos showed:

• *tests/ case-insensitive dir suffix: adds only 1 pp on the weighted total (the file globs already catch the same files by a different path), while carrying real false-positive risk on unrelated PascalCase dirs (Contests/, Requests/, Interests/, Manifests/).
• uitests exact rule: zero incremental hits — UI-test dirs' contents are already *UITests.swift, caught by the file glob.

Token impact (measured across 20 public Swift repos)

Methodology: gh api .../git/trees?recursive=1, Swift source = .swift, 1 MiB ≈ 262K tokens.

Project	Before (analyzed)	After (analyzed)	Reduction
stripe/stripe-ios	14.29 MB / ~3.57M tok	9.47 MB / ~2.37M tok	−1.20M (34%)
kickstarter/ios-oss	10.38 MB / ~2.59M tok	7.12 MB / ~1.78M tok	−0.81M (31%)

• stripe-ios — highest %, 605 file-glob hits. <Framework>Tests/ sub-projects (StripeIdentityTests, StripePaymentSheetTests) each contain hundreds of inline *Tests.swift files.
• kickstarter/ios-oss — 461 file-glob hits. Layered project colocates Foo.swift + FooTests.swift throughout Library/, KsApi/.

Weighted total across the 10-repo Xcode-app subset (75 MB): −2.57M tok / 14%. Range across the full 20-repo sample: 0% (SPM libraries whose Tests/ is already caught) to 34% (Xcode consumer apps).

Why Swift, why these patterns

Consumer iOS apps are where these patterns earn their weight. SPM libraries incur zero behaviour change under the opt-in model. This is the same "layout the existing rules miss" story that showed up for C++ (nlohmann/json 62% in #480), Ruby (rubocop 67% in #597), Python (tensorflow 47% in #579), and Rust (parity-common 28% in #582).

Out of scope

• <AppName>Tests/ / <AppName>UITests/ Xcode dir handling (measurement rejected — marginal 1 pp with false-positive risk).
• Mock/stub file patterns (*Mock.swift, Fake*.swift) — potential follow-up if false-positive risk can be controlled.
• PHP / Dart / C / Lua groups — follow-ups tracked separately.