[CORE-16405] cl/test: uncap target consumer reads for compacted syncs

[andrwng]

The target consumer carried a fixed read-count cap (max_msgs=msg_count) while the source consumer did not. Under compaction, workload completion is gated on per-partition offset parity (max_offsets_match), which a bounded read cannot reach: rebalance re-reads spend the budget, and a partition whose leadership churns has its tail produced and replicated late, so the consumer stops below max_offsets_produced and the workload never finishes reading to the HWM despite replication finishing.

This commit drops the cap on the compacted path so the target consumer tails to parity like the source consumer; the non-compacted path keeps its count-based cap unchanged.

To avoid this in the future, we also now reject max_msgs with use_compaction up front so the incompatibility fails fast instead of stalling for the progress timeout.

This attempts to fix ShadowLinkingRandomOpsTest.test_node_operations, which is quite flaky.

Backports Required

• none - not a bug fix
• none - this is a backport
• none - issue does not exist in previous branches
• none - papercut/not impactful enough to backport
• v26.1.x
• v25.3.x
• v25.2.x

Release Notes

• None

[Copilot]

Pull request overview

Adjusts the cluster linking workload verifier’s target consumer behavior under compaction so it can read until per-partition offset parity (instead of stopping early due to a message-count cap), addressing hangs/flakiness when compaction and rebalances/leadership churn occur.

Changes:

• Add a guard rejecting consumer_properties["max_msgs"] when use_compaction=True to fail fast instead of stalling.
• Remove the target consumer’s fixed max_msgs=self.msg_count cap when use_compaction=True (keep the cap for the non-compacted path).

Comments suppressed due to low confidence (1)

tests/rptest/tests/cluster_linking_test_base.py:299

• max_msgs is passed explicitly to KgoVerifierConsumerGroupConsumer and **self.consumer_properties is expanded into the same call. If consumer_properties ever includes max_msgs (even None), this will raise TypeError: got multiple values for keyword argument 'max_msgs'. Consider copying and popping max_msgs out of the properties before expanding them.

            group_name=f"target-cg-{self._instance_id}",
            nodes=self.preallocated_nodes,
            continuous=True,
            **self.consumer_properties,
        )

[andrwng]

/ci-repeat 5
skip-redpanda-build
skip-units
skip-rebase
dt-repeat=10
tests/rptest/tests/shadow_linking_rnot_test.py::ShadowLinkingRandomOpsTest.test_node_operations