Modified BlockStore trait

[organizedgrime]

Here, many changes have been made, but all within the confines of BlockStores and how they operate.

BlockStores are no longer passed around mutably. Instead, an individual implementation of a given BlockStore will use RefCells, Locks, Cows, etc. to achieve interior mutability in cases where it is required. This changes the signature and lifetimes for many functions, including the put_block and get_block functions required by the trait itself.

[codecov]

Codecov Report

Merging #257 (6988855) into main (020aac2) will decrease coverage by 2.81%.
The diff coverage is 21.19%.

❗ Current head 6988855 differs from pull request most recent head 2bdb463. Consider uploading reports for the commit 2bdb463 to get more accurate results

@@            Coverage Diff             @@
##             main     #257      +/-   ##
==========================================
- Coverage   60.06%   57.26%   -2.81%     
==========================================
  Files          38       40       +2     
  Lines        2697     2885     +188     
  Branches      666      712      +46     
==========================================
+ Hits         1620     1652      +32     
- Misses        624      769     +145     
- Partials      453      464      +11     

Impacted Files	Coverage Δ
wnfs-common/src/blockstore/carblockstore.rs	0.00% <0.00%> (ø)
wnfs-common/src/blockstore/diskblockstore.rs	0.00% <0.00%> (ø)
...ommon/src/blockstore/threadsafememoryblockstore.rs	0.00% <0.00%> (ø)
wnfs-hamt/src/hamt.rs	20.58% <0.00%> (ø)
wnfs-hamt/src/node.rs	55.03% <0.00%> (ø)
wnfs-hamt/src/pointer.rs	30.00% <0.00%> (ø)
wnfs/examples/private.rs	0.00% <0.00%> (ø)
wnfs/examples/tiered_blockstores.rs	0.00% <0.00%> (ø)
wnfs/src/private/link.rs	47.29% <ø> (-1.36%)	⬇️
wnfs/src/private/node/header.rs	86.66% <0.00%> (+1.18%)	⬆️
... and 17 more

... and 8 files with indirect coverage changes

[appcypher]

Another great PR. Thank you for creating this.
We've wanted change the ref mut blockstore params to ref so this partly fixes #251.
I have some comments on the PR. They are mostly nits.

[appcypher]

Another great PR. Thank you for creating this.
We've wanted change the ref mut blockstore params to ref so this partly fixes #251.
I have some comments on the PR. They are mostly nits.

[matheus23]

Thanks for this Vera!

The BlockStore trait adjustment seems good :)
create_cid looks fine to me for now. But for reference and because it's interesting, other blockstore trait implementations go even further: They split out a function put_keyed which takes the "pre-baked" Cid as a parameter.

---

About where the blockstore implementations belong: IMO it makes most sense to put them closer to your data prep tool, outside of rs-wnfs.
We want to keep rs-wnfs' dependencies pretty slim, so we can keep compiling it to wasm (where we'd need different block store implementations). That's also the reason we've kept the blockstore implementations we are using outside of rs-wnfs (except for the one used for tests).
Blockstores are actually a more abstract concept that doesn't have to be used to store wnfs data. The only reason we came up with our own trait is that there wasn't an async variant of a blockstore trait out there yet. The plan is to eventually move the trait outside of rs-wnfs into a common lib that we & e.g. noosphere are using.

[matheus23]

Just some small nits remaining, but otherwise GTG!

---

Hm. This PR has merge conflicts now that #265 landed. I know it can be annoying (to have to repeatedly re-rebase a refactoring PR while smaller PRs are coming in). If you want to, we can take over rebasing @organizedgrime, just let us know (and enable the "allow edits from maintainers" thingie on this PR unless you already have).