Loki config fix

Loki seems to immediately fall over itself given default configurations.
Too many requests due to too much sharding/parallelization? Hard for me
to say. In any case, by changing `parallelise_shardable_queries`, and by
making `max_outstanding_per_tenant` and `split_queries_by_interval`
settings explicit (because in theory these are already the default
settings...), I hope to acheive a more robust service.

Logging stack

syslog-ng -> promtail -> loki -> grafana

A new tag for haproxy which supports syslog logging with practically no
configuration required.