Serialize seek history

[laomaiweng]

Your checklist for this pull request

• I've read the guidelines for contributing to this repository
• I made sure to follow the project's coding style
• I've documented or updated the documentation of every function and struct this PR changes. If not so I've explained why.
• I've added tests that prove my fix is effective or that my feature works (if possible)
• I've updated the rizin book with the relevant information (if needed)

Detailed description

This PR implements seek history serialization/deserialization to/from the Rizin project file.

This is something that I feel is worth preserving across sessions, as seek history is often the logical path the user took to reach some location within the binary. I know in my case I often use it as a lifeline to go back up when I go down some rabbit hole in a binary, and having that lifeline lost if I close Rizin is annoying, hence the PR.

Since I'm new to this code base, there are a number of implementation details that I feel I should mention:

• SDB format
  • the seek history items are serialized as SDB entries with the item index as the entry key, which means the entries
    • may not be ordered numerically by key in the SDB (I figured that doesn't really matter)
    • have to be sorted numerically upon load before parsing the items into the seek history
  • this could have been avoided by storing the items as a JSON array but that felt less legible
  • also using the SDB entry key for a numerical sort means the sort will behave badly if (for some reason, e.g. hand-crafted project file) some key isn't a number, but I figured if the project file is malformed it's not my problem as long as rizin doesn't crash
• current offset
  • the current seek offset is saved as part of the seek history (with key 0 & "is_current": true), which is somewhat redundant with the core/offset entry (not fully redundant since the current seek history item also holds the cursor position which is otherwise not saved)
  • for now the offset of the current item in the seek history is just ignored upon load, I figured at some point in the future it would probably make sense to drop core/offset and use exclusively the offset from the current seek history item
• I added a new vector API for swapping two items in a vector, but it's only used by the seek history deserialization code so if that's not enough users for it to be an API just yet I can remove it

Test plan

Green CI, incl. added unit tests (for seek history serialization & the new rz_vector_swap API) and integration tests (for the bumped project version).

Closing issues

N/A

[laomaiweng]

Updated the db/cmd/project test to account for the project version bump.

[wargio]

Looks good, just some small details.

[laomaiweng]

Fixed these, thanks for the review!

[laomaiweng]

Update for the rz-bindgen lint check failure.

[wargio]

LGTM

[laomaiweng]

CI fails on MSVC because it doesn't support VLAs (in rz_vector_swap), can I #ifdef my way around it and use alloca instead on MSVC only ?

[ret2libc]

why s_value instead of u_value here?

[laomaiweng]

Because cursor in RzCoreSeekItem is an int, so I assumed it may be negative sometimes?

[laomaiweng]

Update for typo & MSVC build failure.

[thestr4ng3r]

Overall a sensible addition and good code!

[laomaiweng]

Updated with the suggested changes, thanks for the review!
I realize I forgot to even include the rationale for the change in the PR, I'll edit it to add that.

[laomaiweng]

Not sure what's wrong with the failed db/formats/dmp/dmp test, there's only been a single failure in CI and I can't reproduce locally. Perhaps a network error? It seems to be fetching a PDB.