TypeSafeConstant siblings: use Map.computeIfAbsent in getConstant

[Vest]

SonarLint java:S3824 cleanup. Three sibling TypeSafeConstant classes (Type, Region, DisplayLocation) had identical get/null-check/put bodies in their getConstant factories:

// before
public static Type getConstant(String name) {
    Type type = TYPE_MAP.get(name);
    if (type == null) {
        type = new Type(name);
        TYPE_MAP.put(name, type);
    }
    return type;
}

// after
public static Type getConstant(String name) {
    return TYPE_MAP.computeIfAbsent(name, k -> new Type(name));
}

Same for Region.getConstant (preserving the leading initializeTypeMap() call) and DisplayLocation.getConstant.

Why the lambda captures name, not k

CaseInsensitiveMap.computeIfAbsent overrides the JDK default to pass resolveObject(key) — a fresh CaseInsensitiveString(key) wrapper — into the mapping function. The function's parameter type is therefore ? super Object, not ? super String. A Type::new method reference would not satisfy that signature.

The lambda captures the outer name parameter (a String) directly, which is exactly what each constructor expects. Behaviour is identical to the prior code: CaseInsensitiveString.toString() returns its underlying string, so even if the lambda used k.toString() it would produce the same value.

What's NOT changed

The sibling lookup-or-throw methods in Region.valueOf and DisplayLocation.valueOf keep their get(name); if null throw shape — that's the intentionally-different "fail-on-missing" idiom, not a candidate for computeIfAbsent.

Verification

• ./gradlew compileJava — BUILD SUCCESSFUL
• ./gradlew :test --tests "pcgen.cdom.enumeration.*" --tests "pcgen.cdom.facet.*" --tests "pcgen.core.*" --tests "plugin.lsttokens.*" — 12 241 tests, 0 failures
• SpotBugs total unchanged at 77 (this is not a SpotBugs PR; the check is for regression sanity)

Diff size

3 files changed, +3 / −21 lines.

Scope note

Independent of all the in-flight SpotBugs PRs (#7631–#7639). Sits on top of fresh origin/master.