Improve ID generation scheme

[ml-evs]

Blocked by #54

Addresses suggestions in #53:

• More ergonomic IDs generation
• immutable_id used to store current path-based ID format
• add last modified based on last rendering of API

[ml-evs]

Hi @eimrek, think this is ready for review whenever you are. Probably easier to start from the changes to the tests and make sure you agree with the idea (as discussed in #53) -- happy to make any changes you see fit!

[eimrek]

Sorry that i didn't have time before. Took a look now, looks great.

But, for the example of let's say

structures.zip:
  structures
  │   ├── molecules
  │   │   ├── benzene.xyz
  │   │   └── butadiene.xyz
  │   ├── set1
  │   │   ├── 55c564f6-ac6a-4122-b8d9-0ad9fe61e961.cif
  │   │   ├── 991bec7a-b3a8-49af-ba6d-be5afd685cd4.cif
  │   │   └── cc1a41b1-a841-4818-baf1-a6c1441dc52a.cif
  │   └── set2
  │       ├── 2aee2118-f484-4d53-90dc-0464cc6c0995.cif
  │       ├── 41b8077b-e051-45ed-88b4-57d8ecce59d0.cif
  │       └── 8b4cac08-928f-443a-8a26-5042d9970d76.cif

I think I would want to keep the subfolder as well (molecules, set1, set2) and possibly the extension, if they're different for different files. I think the IDs could be, in this case,

molecules/benzene.xyz
molecules/butadiene.xyz
set1/55c564f6-ac6a-4122-b8d9-0ad9fe61e961.cif
set1/991bec7a-b3a8-49af-ba6d-be5afd685cd4.cif
set1/cc1a41b1-a841-4818-baf1-a6c1441dc52a.cif
set2/2aee2118-f484-4d53-90dc-0464cc6c0995.cif
set2/41b8077b-e051-45ed-88b4-57d8ecce59d0.cif
set2/8b4cac08-928f-443a-8a26-5042d9970d76.cif

what do you think?

With the current implementation, i get the subfolders and extensions removed.

What i propose is to just remove any common folders/files from front or back of the path (based on splitting on forwardslash). and potentially also removing extension, if common for all the files. Do you see any issues with this?