sled-agent: Change timeout waiting to find switch zones

[jgallagher]

sled-agent attempts to find the IPs of all switch zones in two places:

1. During rack setup (to know what to tell Nexus during handoff)
2. When starting a zone that requires external connectivity (to set up NAT entries in dendrite)

Prior to this PR, both of these would attempt to find both switches for 5 minutes, and after that point would proceed as long as they'd found at least one. This is sorta-okay-but-not-really (more on this in another issue shortly) fine for NAT entries, because Nexus has a background task that will come back around and sync NAT entries for services eventually. But it's not fine for rack setup: if we proceed with only one switch found when the RSS config specifies uplinks for both, we'll fail to hand off to Nexus (details in #9678).

After this change, we change rack setup to wait forever for all switches which have a configured uplink. This means if a switch hasn't come up yet RSS won't proceed, but that should be okay. (It seems better if we could come up with one switch then have Nexus reconcile things after the fact, but that will be a larger change with more risk and more testing difficulty, I think.)

Fixes #9678.

[jgallagher]

Testing notes from dublin: I manually shut off one of the switch zones and kicked off RSS. (This reproduced #9678 prior to this change.)

RSS got stuck, as expected, waiting to find both switch zones:

00:16:10.642Z WARN SledAgent (RSS): Failed to look up switch zone locations
    error = Only found one switch (expected two): {Switch0: fd00:1122:3344:101::2}
    file = sled-agent/src/bootstrap/early_networking.rs:195
    requested_wait_time = 18446744073709551615.999999999s
    retry_after = 17.88171432s
    total_elapsed = 539.955145719s
...
00:16:33.820Z WARN SledAgent (RSS): Failed to look up switch zone locations
    error = Only found one switch (expected two): {Switch0: fd00:1122:3344:101::2}
    file = sled-agent/src/bootstrap/early_networking.rs:195
    requested_wait_time = 18446744073709551615.999999999s
    retry_after = 19.06689036s
    total_elapsed = 563.133314023s
...
00:16:52.959Z WARN SledAgent (RSS): Failed to look up switch zone locations
    error = Only found one switch (expected two): {Switch0: fd00:1122:3344:101::2}
    file = sled-agent/src/bootstrap/early_networking.rs:195
    requested_wait_time = 18446744073709551615.999999999s
    retry_after = 18.115925884s
    total_elapsed = 582.272329612s

Once I restarted the missing switch zone, we found it and proceeded with RSS:

00:17:11.140Z INFO SledAgent (RSS): Successfully looked up all expected switch zone underlay addresses
    addrs = {Switch0: fd00:1122:3344:101::2, Switch1: fd00:1122:3344:103::2}
    file = sled-agent/src/bootstrap/early_networking.rs:159

and successfully handed off to Nexus

19:31:49.938Z INFO SledAgent (RSS): Handing off control to Nexus
    file = sled-agent/src/rack_setup/service.rs:778
19:31:52.640Z INFO SledAgent (RSS): Handoff to Nexus is complete
    file = sled-agent/src/rack_setup/service.rs:1076