[ISSUE-358] Implement ISO-GQL labeled predicate

[yyyCode]

What changes were proposed in this pull request?

Closes #358.

Implements the ISO-GQL <labeled predicate> (ISO/IEC 39075 Section 19.9), which tests whether a graph element (vertex or edge) carries a given label.

Following the pattern of the already-merged source/destination predicate (#366 / PR #675), this is implemented as built-in UDFs rather than a grammar change:

• IS_LABELED(element, label) — returns TRUE if the vertex/edge has the given label, FALSE otherwise.
• IS_NOT_LABELED(element, label) — the negation.

Both follow ISO-GQL three-valued logic: a null element or null label yields Unknown (null). A non-graph-element first operand raises a clear IllegalArgumentException.

Files:

• LabeledPredicateFunctions (geaflow-dsl-common) — shared static logic.
• IsLabeled / IsNotLabeled (geaflow-dsl-plan) — UDFs delegating to the logic, registered in BuildInSqlFunctionTable.

Example:

MATCH (a) -[e]-> (b)
WHERE IS_LABELED(a, 'person')
RETURN a, e, b

How was this PR tested?

• Tests have Added for the changes

• Production environment verified

• LabeledPredicateFunctionsTest (geaflow-dsl-common): unit-tests the logic directly — vertex/edge label matching, negation, three-valued null handling, and rejection of non-graph-element operands. Passes 5/5 locally.

• GQLLabeledPredicateTest (geaflow-dsl-runtime): an end-to-end .sql/.txt case over the modern graph, exercising IS_LABELED/IS_NOT_LABELED on both vertices and edges. Expected outputs were computed by hand from the graph data (not copied from program output).

mvn checkstyle:check reports 0 violations across the three touched modules.

Note: the runtime end-to-end test could not be executed on my local Windows machine because Hadoop's winutils.exe is required to write the file sink (the existing GQLSourceDestinationTest fails locally for the same reason). It is expected to run on the Linux CI. The predicate logic itself is fully covered by the platform-independent unit test.

[yaozhongq]

This package is intended for general-purpose interfaces. There is no need to split the current function into a separate Func; I suggest placing it directly into the UDF.

[yyyCode]

Good point, thanks. Fixed in 5c65fc8: I moved the label-matching and three-valued-null logic (plus a private getLabel helper) directly into IsLabeled, and IsNotLabeled now delegates to an IsLabeled instance and negates the result, so label extraction stays in one place. Removed LabeledPredicateFunctions and its test from geaflow-dsl-common, and moved the unit test into geaflow-dsl-plan as LabeledPredicateTest (same udf/table/other package as the UDFs, matching PropertyExistsTest), testing through the UDF eval methods.

Note: the existing merged predicates (IsSourceOf/PropertyExists) still use the split helper pattern this PR was originally modeled on, so the labeled predicate is now self-contained and diverges from them. Happy to align those too in a follow-up if you would prefer consistency.

[yyyCode]

Thanks for the review @yaozhongq. Confirmed on the latest commit: LabeledPredicateFunctions is gone from geaflow-dsl-common, the label-matching + three-valued-null logic (and the getLabel helper) now lives directly in IsLabeled, and IsNotLabeled delegates to an IsLabeled instance so label extraction stays in one place. The unit test moved to geaflow-dsl-plan as LabeledPredicateTest (same package as the UDFs, matching PropertyExistsTest) and passes 5/5 locally. PTAL when you get a chance.

[yaozhongq]

LGTM, looks great!

[yaozhongq]

LGTM