header cache is still not safe

[brendan]

The header cache does not properly dump or zero several fields in dump_body, leaving several pointers behind that are likely to be invalid. At first glance, the bad fields seem to include charset, content, next, parts, hdr, and aptr, unless these are reset elsewhere in the code after restore_body.

[pdmef]

Add patch updating hcache upon mailbox sync for IMAP, Maildir and MH (can't do it for POP as the ACLs we need to enable for updating to be useful would allow too many operations like editing messages).

[pdmef]

(In [b9ac445b035b]) Prevent some pointers of 'struct body' being saved to hcache

This addresses the hcache safety issue but maybe doesn't fix it completely, see #2942. This also prevents mutt from crashing when using a hcache that was synced when syncing the mailbox (updating changed and fully-parsed messages).