Re-enable 3 flaky multi-client session resume tests via isolated Surefire execution

The tests were @disabled in PR 171 due to timeouts, but pass reliably in
isolation. Rather than leaving them disabled, use a JUnit 5 @tag and a
separate Surefire execution to run them in their own forked JVM, avoiding
harness state leakage from the main test suite.

pom.xml:
  - Add <executions> block to maven-surefire-plugin with two executions:
    1. "default-test": runs all tests EXCEPT those tagged "isolated-resume"
    2. "isolated-resume-tests": runs ONLY tests tagged "isolated-resume"
  - Each execution forks a separate JVM, providing process-level isolation.

src/test/java/com/github/copilot/sdk/CopilotSessionTest.java:
  - Remove @disabled annotation from testShouldResumeSessionUsingNewClient
  - Add @tag("isolated-resume") to the test method
  - Replace 'import Disabled' with 'import Tag'

src/test/java/com/github/copilot/sdk/StreamingFidelityTest.java:
  - Remove @disabled from testShouldProduceDeltasAfterSessionResume
  - Remove @disabled from testShouldNotProduceDeltasAfterSessionResumeWithStreamingDisabled
  - Add @tag("isolated-resume") to both test methods
  - Replace 'import Disabled' with 'import Tag'

Author: edburns

pom.xml

@@ -327,9 +327,8 @@
                         This handles intermittent failures in E2E tests (e.g.,
                         CompactionTest) where snapshot matching can be sensitive
                         to non-deterministic compaction behaviour.
-                        TODO: Fix the root cause of CompactionTest flakiness
-                        (snapshot mismatch after compaction) so this retry is
-                        no longer needed.
+                        Revisit this once this issue is successfully resolved.
+                        https://github.com/github/copilot-sdk/issues/1227 
                     -->
                     <rerunFailingTestsCount>2</rerunFailingTestsCount>
                     <systemPropertyVariables>
@@ -346,6 +345,30 @@
                         <COPILOT_CLI_PATH>${copilot.cli.path}</COPILOT_CLI_PATH>
                     </environmentVariables>
                 </configuration>
+                <executions>
+                    <!--
+                        Run multi-client session resume tests in isolation BEFORE the
+                        main suite. These tests pass reliably alone but can time out
+                        when run after other E2E tests due to harness state leakage.
+                    -->
+                    <execution>
+                        <id>isolated-resume-tests</id>
+                        <phase>test</phase>
+                        <goals>
+                            <goal>test</goal>
+                        </goals>
+                        <configuration>
+                            <groups>isolated-resume</groups>
+                        </configuration>
+                    </execution>
+                    <!-- Exclude the isolated resume tests from the main run -->
+                    <execution>
+                        <id>default-test</id>
+                        <configuration>
+                            <excludedGroups>isolated-resume</excludedGroups>
+                        </configuration>
+                    </execution>
+                </executions>
             </plugin>
             <!-- Add src/generated/java as an additional source root -->
             <plugin>

src/test/java/com/github/copilot/sdk/CopilotSessionTest.java

@@ -20,7 +20,7 @@
 
 import org.junit.jupiter.api.AfterAll;
 import org.junit.jupiter.api.BeforeAll;
-import org.junit.jupiter.api.Disabled;
+import org.junit.jupiter.api.Tag;
 import org.junit.jupiter.api.Test;
 
 import com.github.copilot.sdk.generated.SessionEvent;
@@ -307,7 +307,7 @@ void testShouldResumeSessionUsingTheSameClient() throws Exception {
      * @see Snapshot: session/should_resume_a_session_using_a_new_client
      */
     @Test
-    @Disabled("Passes in isolation but times out in full suite due to test interaction (state leakage or process contention)")
+    @Tag("isolated-resume")
     void testShouldResumeSessionUsingNewClient() throws Exception {
         ctx.configureForTest("session", "should_resume_a_session_using_a_new_client");
 

src/test/java/com/github/copilot/sdk/StreamingFidelityTest.java

@@ -14,7 +14,7 @@
 
 import org.junit.jupiter.api.AfterAll;
 import org.junit.jupiter.api.BeforeAll;
-import org.junit.jupiter.api.Disabled;
+import org.junit.jupiter.api.Tag;
 import org.junit.jupiter.api.Test;
 
 import com.github.copilot.sdk.generated.SessionEvent;
@@ -140,7 +140,7 @@ void testShouldNotProduceDeltasWhenStreamingIsDisabled() throws Exception {
      * @see Snapshot: streaming_fidelity/should_produce_deltas_after_session_resume
      */
     @Test
-    @Disabled("Passes in isolation but times out in full suite due to test interaction (state leakage or process contention)")
+    @Tag("isolated-resume")
     void testShouldProduceDeltasAfterSessionResume() throws Exception {
         ctx.configureForTest("streaming_fidelity", "should_produce_deltas_after_session_resume");
 
@@ -194,7 +194,7 @@ void testShouldProduceDeltasAfterSessionResume() throws Exception {
      *      streaming_fidelity/should_not_produce_deltas_after_session_resume_with_streaming_disabled
      */
     @Test
-    @Disabled("Passes in isolation but times out in full suite due to test interaction (state leakage or process contention)")
+    @Tag("isolated-resume")
     void testShouldNotProduceDeltasAfterSessionResumeWithStreamingDisabled() throws Exception {
         ctx.configureForTest("streaming_fidelity",
                 "should_not_produce_deltas_after_session_resume_with_streaming_disabled");