Update driver tests for opportunistic rounds/portRounds tracking

[Copilot]

The collectTask/feedInTask progress tracking was refactored from a single round barrier counter to per-port portRounds, enabling opportunistic processing. The existing unit tests still referenced the old round-based model and needed to be updated.

Changes

• Test assertions: Updated FeedIn/Collect API tests to assert on rounds and portRounds fields instead of the removed round field
• Feed-in task setup: Updated manual feedInTask construction in "should do feed in" to use rounds/portRounds
• Opportunistic tests: Added "should do feed in opportunistically" and "should do collect opportunistically" — these verify per-port independent progress when individual ports stall:

localPort2.EXPECT().CanSend().Return(false).Times(1)
localPort2.EXPECT().CanSend().Return(true).AnyTimes()
// After tick 1: port 2 is still at round 0, ports 1&3 advanced
Expect(task.portRounds).To(Equal([]int{1, 0, 1}))

• Validation tests: Added "should reject invalid FeedIn/Collect layout" tests covering the new validateTaskLayout panic paths (non-multiple data length, stride < port count)

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.