Use intersphinx links for links to the CrateDB docs

Sphinx has an extension called "intersphinx"
(https://www.sphinx-doc.org/en/3.x/usage/extensions/intersphinx.html)
which allows cross-referencing multiple Sphinx projects. By relying on
the extension, dead links can be detected earlier. Additionally, it adds
the same semantics and contextual details for links between
documentation projects. For example, a link from the CrateDB Kubernetes
Operator docs to something in its code, would look like this:

  :func:`crate.operator.main.create`

Linking to a function in another project, listed in the
`intersphinx_mapping` could be

  :func:`kopf:kopf.on.create`

In both cases, Sphinx will know that it's linking to a function, thus
using the correct visual attributes when rendering the HTML. Unlike
linking to the URL directly, which would only show as a regular link.

docs/source/concepts.rst

@@ -48,9 +48,9 @@ Operator will perform a rolling cluster restart, following this process:
 
 Restarting a CrateDB node works by deleting the corresponding Kubernetes pod.
 The operator will then wait for a new pod to go down and another one to come up
-and join the CrateDB cluster. Finally, the cluster needs to `report a green
-health label`_ three times in a row, with a 30 second wait time in between, for
-a pod to be considered back in service.
+and join the CrateDB cluster. Finally, the cluster needs to :ref:`report a
+green health label <cratedb:sys-health>` three times in a row, with a 30 second
+wait time in between, for a pod to be considered back in service.
 
 
 Cluster Scaling
@@ -92,8 +92,8 @@ The scaling operation will follow these four basic steps:
 #. The last step for the operator is to update all StatefulSets that make a
    CrateDB cluster to the expected total number of nodes. This ensures that if
    a pod dies, its restart won't alter CrateDB's expected number of nodes. This
-   will also involve acknowledging all ``gateway.expected_nodes`` `node
-   checks`_ (ID 1).
+   will also involve acknowledging all ``gateway.expected_nodes`` :ref:`node
+   checks <cratedb:sys-node-checks>` (ID 1).
 
 The entire scaling operation may not take longer than 3600 seconds by default,
 or no longer than whatever is configured in the :envvar:`SCALING_TIMEOUT`
@@ -186,7 +186,3 @@ documented below.
    :``new_version``:
       The new image version (Docker tag) as defined in
       ``.spec.cluster.version``.
-
-
-.. _report a green health label: https://crate.io/docs/crate/reference/en/latest/admin/system-information.html#health
-.. _node checks: https://crate.io/docs/crate/reference/en/latest/admin/system-information.html#node-checks