Summary

Follow-up to #2943, adopting the cleaner design @Naenyn proposed. Replaces the @SuppressWarnings/unused workaround that just landed with the correct type for a fire-and-forget callback: java.util.function.Consumer<T>. With Consumer, .accept() returns void and the ErrorProne CheckReturnValue check doesn't apply at all — no suppression needed.

Context — why this is a follow-up rather than an update to #2943

@Naenyn had pushed his Consumer-based version (Naenyn/uPortal@f2e3a459) to his remove_fluid branch and asked me to adopt it. I force-pushed the cherry-pick onto #2943's branch, but the merge of #2943 had already resolved against the original @SuppressWarnings version — the better approach didn't make it in.

This PR lands the same set of changes as a clean follow-up, with @Naenyn credited as co-author. When his #2915 rebases master, these hunks will be no-ops (already on master) and his PR shrinks correspondingly.

Scope (6 files, +25/−56)

Source (uPortal-rendering):

• LimitingTeeWriter.java — Function<LimitingTeeWriter, ?> → Consumer<LimitingTeeWriter>, .apply(this) → .accept(this), swap import, update Javadoc reference.
• LimitingTeeOutputStream.java — same pattern.
• CachingPortletOutputHandler.java (sole caller) — anonymous Function<> classes → concise lambdas:

  input -> clearCachedWriter()
  input -> clearCachedStream()

  Drop the now-unused com.google.common.base.Function import.

Tests (uPortal-webapp):

• LimitingTeeWriterTest.java, LimitingTeeOutputStreamTest.java — matching Function-to-lambda updates; drop unused Function imports.
• PortalRawEventsAggregatorImplTest.java — different pattern here: the Function comes from the code-under-test's own API, not our types, so it can't be changed to Consumer. Use Boolean ignored = … to capture the return (same shape @Naenyn used; aligns with the ErrorProne convention for the cases where the type genuinely can't be changed).

Validation

• ./gradlew :uPortal-rendering:compileJava → BUILD SUCCESSFUL
• ./gradlew :uPortal-webapp:compileTestJava → BUILD SUCCESSFUL
• @Naenyn's remove_fluid branch (which has resourceServerVersion=1.5.2 and therefore triggers the underlying Guava @CheckReturnValue check) already builds clean with these exact changes — so the fix addresses the real failure mode.

Net effect

After this lands, master will carry @Naenyn's original Consumer design rather than the @SuppressWarnings workaround. Cleaner code, no behavior change.

cc @Naenyn