srv_topo_timeout of 1s too aggressive

[aquarapid]

A recent change (#7278), introduced a timeout for srv_topo operations. The default value was set at 1 second. This seems reasonable, but it has had the unintended side-effect of causing topo timeouts, especially during vtgate startup (tablet wait/healthcheck), for users that have topology spread across multiple DCs.

In the interest of sane defaults, I suggest we lift the default to 5 seconds; and those who need a lower value can set it down.

For discoverability, here is the type of error you would see:

E0430 16:34:01.359471       1 resilient_server.go:311] GetSrvKeyspaceNames(context.Background.WithDeadline(2021-04-30 16:34:30.359023101 +0000 UTC m=+30.040465073 [28.999518338s]), xxxx4a1) failed: deadline exceeded: /vitess/tst006/global/cells/xxxx4a1/CellInfo (no cached value, caching and returning error)
F0430 16:34:01.359543       1 vtgate.go:157] gateway.WaitForTablets failed: deadline exceeded: /vitess/tst006/global/cells/xxxx4a1/CellInfo

[deepthi]

Fixed by #8011