e2e: wait for multicast-group convergence in doublezero status check

[ben-dz]

Summary of Changes

• Wrap the post-connect doublezero status assertion in checkMulticastPostConnect in a polling loop so it waits for multicast-group membership to converge instead of taking a single, non-retried snapshot. This fixes a test-side race where the second group (mg02) subscription had not yet propagated into the client daemon's local status view, producing S:mg01 instead of the expected S:mg01,S:mg02 and failing the CLI-table diff.
• The fix is shared by both the publisher and subscriber paths. The sibling doublezero_multicast_group_list check is intentionally left as a single snapshot because it reads onchain state, which is already consistent.
• The polling timeout/interval (60s/1s) matches the existing convergence-sensitive checks in the same file (e.g. the multicast route check).
• On a genuine timeout, the last observed table and structured -(want), +(got) diff (or the last exec error) are still printed, so a real failure is not opaque.
• This is a pre-existing race that happened to surface on PR build(deps): bump openssl from 0.10.73 to 0.10.81 #3900; it is fixed here and not bundled into that PR.
• Fixes e2e: flaky TestE2E_Multicast subscriber status check races multicast-group convergence #3907

Testing Verification

• go vet -tags e2e ./e2e passes; the package compiles and is gofmt-clean.
• The race is timing-dependent and only reproduces under load across repeated full TestE2E_Multicast runs (each spins up multiple cEOS containers), so it cannot be deterministically reproduced in a single run. The change converts the single-snapshot assertion that lost the race into the same Eventually-poll idiom already proven for the route/PIM convergence checks in this file, which read the same asynchronously-populated daemon state.