e2e: increase waitForRoutesTimeout from 90s to 120s

[martinsander00]

Important

The primary goal of this PR is to increase waitForRoutesTimeout from 90s to 120s. The diagnostic test file (qa_bgp_propagation_test.go) is included for reference but will be removed before merging.

Summary

• Increase waitForRoutesTimeout from 90s to 120s to reduce flaky QA test failures
• Add diagnostic test (TestQA_BGPPropagationVariance) used to investigate the root cause

Investigation

QA tests (TestQA_UnicastConnectivity, multicast tests) were intermittently timing out at the "waiting for routes to be installed" step, particularly for the fra↔sgp (Frankfurt↔Singapore) route.

What we tested

1. Manual timing of BGP route propagation between fra-tn-qa01 and sgp-tn-qa01
2. Device comparison: fra-dz001 (jump_ contributor) vs fra-dz-001-x (rox contributor)
3. Variance test: 5 iterations measuring propagation time with detailed timing breakdown

Findings

Iter	Disconnect	Fra Connect (BGP up)	Sgp Connect (BGP up)	Route Propagation	Total
1	5.5s	38.0s	35.7s	0.1s ✨	38.5s
2	7.7s	9.9s	29.8s	48.2s	79.9s
3	7.7s	9.9s	31.3s	45.8s	78.8s
4	12.1s	6.1s	32.3s	35.3s	69.6s
5	11.2s	8.2s	33.3s	33.3s	68.4s

Total = Sgp Connect + Route Propagation (timing starts when sgp begins connecting)

Additional context - Fra initial routes at start of each iteration:

Iter 1	Iter 2	Iter 3	Iter 4	Iter 5
69 routes	0 routes	70 routes	57 routes	70 routes

Key observations:

• Cold vs warm BGP state: When fra has existing routes (warm), new routes propagate instantly (~100ms). When fra's BGP table is empty (cold start), propagation takes 33-48s
• Iteration 2 anomaly: Fra reported "BGP Session Up" but had 0 routes, indicating a gap between session establishment and route exchange
• Total propagation time: sgp connect (~30s) + route wait (33-48s) = 65-80s in worst case
• 90s timeout is borderline: With variance, cold-start scenarios can exceed 90s

Root cause

There's a delay between "BGP Session Up" status and actual route exchange completing. In cold-start scenarios (first test of the day, or after BGP state reset), route propagation between distant exchanges (xfra↔xsin) can take 65-80s total, which approaches or exceeds the 90s timeout.

Solution

Increase waitForRoutesTimeout from 90s to 120s to provide sufficient headroom for worst-case BGP propagation times.

Testing Verification

• Ran TestQA_BGPPropagationVariance 5 iterations - all passed with new 120s timeout
• Ran TestQA_UnicastConnectivity with all 4 hosts - passed in 113s
• go build -tags=qa ./e2e/... passes