Support dumping Turin data structures and Milan data structures at th…

[daym]

…e same time.

Fixes #135.

[daym]

More context see #134 (comment)

General design is right now (all this is just for serde):

• Add context field at the root of the Apcb--visible to the user but not mandatory
• Thread through its value all the way down to entry handlers
• Have the entry Serialize handler check that value and use it to decide which version of MemDfeSearchElement to use.
• Keep the entry Deserialize handler unchanged so that the new context field can be "half" optional (optional for deserialization, optional-ish for serialization). What I mean is that if the Serialize handler finds None in context, it will not pick any of the struct versions and instead serialize out a byte array into JSON.

[luqmana]

Mind up'ing Apcb::MAX_SIZE to 0x8000 or so? Trying to dump turin-1.0.0.0 currently fails:

$ ../../amd-host-image-builder/target/debug/amd-host-image-builder dump -i RRRT1000F.FD -b dump
WARNING: PSP entry with unknown type was skipped PspDirectoryEntry { type_: Err(InvalidBitPattern { invalid_bytes: 105 }), sub_program: Ok(0), rom_id: Ok(SpiCs1), size: Some(8176), source: Ok(PhysicalAddress(0x515500)) }
WARNING: PSP entry with unknown type was skipped PspDirectoryEntry { type_: Err(InvalidBitPattern { invalid_bytes: 105 }), sub_program: Ok(0), rom_id: Ok(SpiCs1), size: Some(8176), source: Ok(PhysicalAddress(0x1fdd00)) }
thread 'main' panicked at src/main.rs:804:53:
range end index 30976 out of range for slice of length 25856
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace