In LDK 0.3 we added support for reconstructing the ChannelManager's pending
forward HTLCs maps from channel data as part of removing the requirement to
regularly persist the manager, but til now we still would write/read those maps
within the manager to maintain compat with 0.2-. Also 0.3 added a new field to
Channel that allowed the map reconstruction.

Now that a few versions have passed we have more confidence that the new field
is being written, here we deprecate persistence of the old legacy maps and only
attempt to read them if the manager serialized version indicates that they were
written.

Upcoming commits will ensure we error if the new field is missing.

XXX clean up claude tests

See previous commit, but now in prod reconstruct_manager_from_monitors will
always be set, so there's no need to have an option to set it in tests.

Used in the next commit when we log on some read errors.

In 0.3, we added a new field to Channel as part of adding support for
reconstructing the ChannelManager's pending forward HTLC set from
Channel{Monitor} data, moving towards removing the requirement of regularly
persisting the ChannelManager.

This new field cannot be set for HTLCs received pre-0.1 that are using this
legacy ::Resolved variant.

In this version, we fully rely on the new Channel field being set and cease
handling legacy HTLCs entirely, and thus must fully deprecate the ::Resolved
variant and require all inbound HTLCs be in state
InboundHTLCResolution::Pending, which we do here. Further cleanups are coming
in upcoming commits.

Previously, several variants of InboundHTLCState contained an
InboundHTLCResolution enum.  Now that the resolution enum only has one variant,
we can get rid of it and simplify the parent InboundHTLCState type.

This version of LDK no longer supports HTLCs created before 0.3, which allows
us to remove this variant that will only be present if an HTLC was received on
a pre-0.3 LDK version. See the past few commits.

In the previous commit, we removed the InboundHTLCResolution::Resolved enum
variant, which caused us to never provide any HTLCs in this now-removed
parameter.

We stopped adding any HTLCs to this vec a few commits ago when we removed
support for HTLCs that were originally received on LDK 0.1.

We stopped adding any HTLCs to this vec a few commits ago when we removed
support for HTLCs that were originally received on LDK 0.1

We stopped using this struct a few commits ago when we removed
support for HTLCs that were originally received on LDK 0.1.