Thanks for the suggestion @jcalderone!

Response time, probe fails and TLS expiry incl. alerts.

The Blackbox exporter can be used to check whether some services do answer;
We'll use it for our HTTPS endpoints.
Especially handy is it's checking for TLS cert expiry.

This change reflects my intention of having an alert fire when a node
is not scrapable for five minutes or longer.  I hadn't understood the
default avg() function well enough before, yielding an alert that would
fire too soon (i.e. any time a node was down, even only one minute for a
reboot).

Fixes privatestorageops#358.

This changes the Throughput graph to be relative, changes its unit
to Bit/s, and fixes the labels of the Network pkt/s graph.

Fixes privatestorageops#24.

Fixes privatestorageops#355.

Epiphany: Systemd reads more than a single environment file. This
lets me improve the file naming much.

Unfortunately still prints a bold warning on deployment.

This will store the supposedly secret URL in the Nix Store and
print a honking big warning about it.

Only dashboards in this dir

... to be in sync with https://github.com/PrivateStorageio/PaymentServer/pull/98

I hadn't understood: The PaymentServer turns 16 vouchers into
50 k ZKAPs.  So far, we don't have a metric for used up ZKAPs.

Also, since the dashboard is already a full browser page and
only handling payment, I rename the services dashboard to payment
dashboard.

Fixes privatestorageops#349.

We have to explicitely add the port number so the relabel
config can strip it out again.  Else, `instance' seems to
get its value from somewhere else - somewhere our relabel
config does not look at.

This commit does not change the semantics, only the looks.

At least it's easy to turn on

This is used in constructing the Google OAuth2 login callback URL and has to
match what's configured in Google - also helps if it is actually the address
of the server.

Fewer lines almost always better.  Also, we don't want to spread
these out unless we have to.