test(agentic-ai): make e2e incident error assertions more resilient (#6452) (#6480)

* test(agentic-ai): make e2e incident error assertions more resilient

Incident error messages may contain additional info, such as error variables. Instead of matching equality, we only match parts of the message.

* test(agentic-ai): fix incident assertion on MCP auth tests not actually asserting the string

(cherry picked from commit 0e43f74)
(cherry picked from commit 8321692)

Author: maff

connectors-e2e-test/connectors-e2e-test-agentic-ai/src/test/java/io/camunda/connector/e2e/agenticai/aiagent/langchain4j/jobworker/L4JAiAgentJobWorkerLimitsTests.java

@@ -100,7 +100,7 @@ private void testMaxModelCallsLoop(
         incident -> {
           assertThat(incident.getElementId()).isEqualTo(AI_AGENT_TASK_ID);
           assertThat(incident.getErrorMessage())
-              .isEqualTo(
+              .startsWith(
                   "Maximum number of model calls reached (modelCalls: %1$d, limit: %1$d)"
                       .formatted(expectedMaxModelCalls));
         });

connectors-e2e-test/connectors-e2e-test-agentic-ai/src/test/java/io/camunda/connector/e2e/agenticai/aiagent/langchain4j/outboundconnector/L4JAiAgentConnectorLimitsTests.java

@@ -88,7 +88,7 @@ private void testMaxModelCallsLoop(
         incident -> {
           assertThat(incident.getElementId()).isEqualTo(AI_AGENT_TASK_ID);
           assertThat(incident.getErrorMessage())
-              .isEqualTo(
+              .startsWith(
                   "Maximum number of model calls reached (modelCalls: %1$d, limit: %1$d)"
                       .formatted(expectedMaxModelCalls));
         });