[MISC]: Bump edition, remove late variable assignments, fix clippy lints, replace new objects with slices, and other.

[Uzaaft]

This pr does a couple of things:

• Fix clippy listing errors
• variable declarations for match/if chains
• change Stringwith &str a few places(This should be done in more places, theoretically it might reduce execution times)
• Bump edition to 2021
  @rob-p I am just wondering. Should I go ahead and bump the version number of both alevin-fry and libradicl?

[rob-p]

@Uzaaft : This is a dependency we have to keep at 0.1.26 for the time being because of a build error on old versions of Clang (that are used in bioconda, upon which we partially rely for distribution). For more details on this see this. Suggest changing this back to 0.1.26 until those upstream changes are merged in.

[Uzaaft]

Fixed in 83e5f9b

[rob-p]

Hi @Uzaaft,

Thanks so much for these changes. For the eventual next tagged release, yes; the version numbers for both of these should change. This is because when we push to crates.io, we need to rely on libradicl in crates rather than the local dev version — so that version will have to be bumped to accomodate a new alevin-fry version, since it has actually changed now.

[rob-p]

By the way, @Uzaaft — these are really nice changes to make the code more idiomatic. Any idea why clippy might not have suggested these by default? Are there just not lints for these yet? Did you notice these by just going through the code by hand?

[Uzaaft]

@rob-p I'm not sure why.
I used cargo clippy in the terminal and clion to highlight the issues. Clion showed a few issues that clippy didn't show, and vice versa.
Although, the &str part was not a clippy lint, just a few hard lessons previously learnt. xD

[Uzaaft]

Which reminds me, Perhaps you(or I xD) should consider moving more usages of String to &str.
If this is something you'd like, I can go ahead and migrate everything I see to &str.
This should speed things up(or keep the performance at the same level).

[rob-p]

That would be great! I guess I don't see the Clion ones because I'm using VSCode and or Doom Emacs for most of the development, so I only get the clippy lints. If you can move over the usages of String to &str that seem reasonable, then I think this PR will be ready to merge.

[Uzaaft]

Emacs? You should try neovim 😝I'll move more usages of string to &str, and tag you when I am done

[rob-p]

To be fair, I only use Doom Emacs with EVIL mode enabled. I can't edit text with non-vim key bindings.

[Uzaaft]

Migrating things to &str is was not as easy as I thought, so it'll take some time.
Also, I see some usage of unsafe. For example line 349 in src/infer.rs.
Just wondering, why do you not want to check of the string is valid utf8?

[rob-p]

That's a good question, @Uzaaft, and the documentation around that unsafe call should be more clear. The reason this is OK is because we control the the file we are ingesting, which contains only valid cell-level tags (here, biological cells, not cell as in RefCell). That is, in the input file, only strings of 2-bit encoded characters from the alphabet {A,C,G,T} are permitted / written. Thus, it should always be safe to assume the string representation is valid UTF8 without checking for that.

[Uzaaft]

@rob-p The migration to &str was too ambitious. I'm going to leave them out for now, and perhaps come back to that during the summer months.
Added a bonus change where we avoid creation of new objects where they aren't needed, instead using slices.
I think the PR is ready to be reviewed.

[rob-p]

Everything here looks great @Uzaaft! Thanks for this PR :), and I'm looking forward to your further suggestions and contributions.

[Uzaaft]

@rob-p I am not sure if this got merged correctly. Just checked the newest master commit, and somehow the changes I made isnt there now

[Uzaaft]

Turns out, I merged the local repo wrong