replication config: retry first three failed times quicker before going to 30m

replication config: retry first three failed times quicker before going to 30m

So the repeat frequency for a stuck job is now:
t0 -> fails
t1 = t0 +  5m -> repat
t2 = t1 + 10m = t0 + 15m -> repat
t3 = t2 + 15m = t0 + 30m -> repat
t4 = t3 + 30m = t0 + 60-> repat
then
tx = tx-1 + 30m -> repat

So, we converge more naturally/stable to the 30m intervals than
before, when t3 would have been t0 + 45m.

Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>