refactor(runtime): align fallback epoch comment and test preamble

Match the file's `// Why:` comment convention on the inherited
publication epoch, and drop a redundant duplicate mocks import in the
lineage regression test while keeping the required side-effect order.

No behavior change.

src/main/runtime/orca-runtime-create-runtime-owned-mobile-session-terminal.ts

@@ -120,7 +120,7 @@ export class OrcaRuntimeWithCreateRuntimeOwnedMobileSessionTerminal extends Orca
     }
     const next: RuntimeMobileSessionTabsSnapshot = {
       worktree: worktreeId,
-      // A fallback adds a surface without retiring the renderer that will publish its next tabs.
+      // Why: a fresh epoch retires the current publisher, so clients drop its later tab updates.
       publicationEpoch: existing?.publicationEpoch ?? `headless:${Date.now().toString(36)}`,
       snapshotVersion: (existing?.snapshotVersion ?? 0) + 1,
       // Why: activating the new tab also focuses its group, so a "+" targeting a specific split group makes that group active too.

src/main/runtime/runtime-owned-terminal-publication-lineage.test.ts

@@ -1,9 +1,9 @@
 import { expect, it, vi } from 'vitest'
 import type { RuntimeMobileSessionTabsResult } from '../../shared/runtime-types'
 
-await import('./orca-runtime-test-mocks.spec')
-await import('./orca-runtime-test-lifecycle.spec')
+// Fragments stay side-effect ordered: mocks, then lifecycle, then fixtures.
 const { OrcaRuntimeService } = await import('./orca-runtime-test-mocks.spec')
+await import('./orca-runtime-test-lifecycle.spec')
 const { store, TEST_WORKTREE_ID } = await import('./orca-runtime-test-fixtures.spec')
 
 it.each(['renderer:active-generation', 'headless:active-generation'])(