Backport of Update backends.mdx into v1.3

Backport of evaluation of missing instances during import into v1.3

Merge pull request #31937 from hashicorp/backport/jbardin/destroy-edge-cycles/formally-neutral-mallard

Backport of Extract more exact provider name when checking for destroy cycles into v1.3

Backport of special handling for legacy `ignore_changes = all` into v1.3

Backport of Update custom-conditions.mdx into v1.3

Backport of Prevent errors from NoOp deposed changes into v1.3

Once again we're caught out by sharing the same output value node type
between the plan phase and the apply phase. To allow for some slight
variation between plan and apply without drastic refactoring here we just
add a new flag to nodeExpandOutput which is true only during the planning
phase.

This then allows us to register the checkable objects only during the
planning phase and not incorrectly re-register them during the apply phase.
It's incorrect to re-register during apply because we carry over the
planned checkable objects from the plan phase into the apply phase so we
can guarantee that the final state will have all of the same checkable
objects that the plan did.

This avoids a panic during the apply phase from the incorrect duplicate
registration.

Merge pull request #31885 from hashicorp/backport/jbardin/ignore-changes-all-computed/centrally-factual-colt

Backport of filter computed attrs from `ignore_changes=all` into v1.3

We use a non-pointer value for this particular node, which means that
there can never be two root nodes in the same graph: the graph
implementation will just coalesce them together when a second one is added.

Our resource expansion code is relying on that coalescing so that it can
subsume together multiple graphs for different modules instances into a
single mega-graph with all instances across all module instances, with
any root nodes coalescing together to produce a single root.

This also updates one of the context tests that exercises resource
expansion so that it will generate multiple resource instance nodes per
module and thus potentially have multiple roots to coalesce together.
However, we aren't currently explicitly validating the return values from
DynamicExpand and so this test doesn't actually fail if the coalescing
doesn't happen. We may choose to validate the DynamicExpand result in a
later commit in order to make it more obvious if future modifications fail
to uphold this invariant.

We previously did two levels of DynamicExpand to go from ConfigResource to
AbsResource and then from AbsResource to AbsResourceInstance.

We'll now do the full expansion from ConfigResource to AbsResourceInstance
in a single DynamicExpand step inside nodeExpandPlannableResource.

The new approach is essentially functionally equivalent to the old except
that it fixes a bug in the previous implementation: we will now call
checkState.ReportCheckableObjects only once for the entire set of
instances for a particular resource, which is what the checkable objects
infrastructure expects so that it can always mention all of the checkable
objects in the check report even if we bail out partway through due to
a downstream error.

This is essentially the same code but now turned into additional methods
on nodeExpandPlannableResource instead of having the extra graph node
type. This has the further advantage of this now being straight-through
code with standard control flow, instead of the unusual inversion of
control we were doing before bouncing in and out of different Execute and
DynamicExpand implementations to get this done.

We were previously _trying_ to handle diagnostics here but were not quite
doing it right because we were testing whether the resulting error was
nil rather than appending it to the diagnostics and then seeing if the
result has errors.

The difference here is important because it allows DynamicExpand to return
warnings without associated errors when needed. Previously the graph
walker would treat a warnings-only result as if it were an error.

Ideally we'd change DynamicExpand to return diagnostics directly, but we
previously decided against that because there were so many implementors
to update, and my intent for this change is to be surgical in the update
so we minimize risk of backporting the change into patch releases.

Merge pull request #31874 from hashicorp/backport/jbardin/remove-planned-during-import/genuinely-shining-boar

Backport of RemovePlannedResourceInstanceObjects during import into v1.3

Backport of prune unused nodes from a destroy plan graph into v1.3

Merge pull request #31872 from hashicorp/backport/jbardin/destroy-edge-cycles/instantly-prepared-hagfish

Backport of prevent cycles when connecting destroy nodes into v1.3

Backport of Update workspace documentation into v1.3