Make types like `Box<dyn DynAccess>` implement Access

[steffahn]

by generalizing the P: Deref-implementation to ?Sized, and implementing Access for some dyn DynAccess trait object types.

Not yet documented or polished, this PRs main purpose for now is to allow CI testing.

[steffahn]

The failing clippy lints are unrelated to this PR.

[steffahn]

Older compiler version apparently fails in one of the dependencies (half).

[codecov-commenter]

Codecov Report

Merging #78 (a113002) into master (f7f192d) will increase coverage by 0.60%. The diff coverage is 33.33%.

@@            Coverage Diff             @@
##           master      #78      +/-   ##
==========================================
+ Coverage   83.37%   83.97%   +0.60%     
==========================================
  Files          18       18              
  Lines        1149     1161      +12     
==========================================
+ Hits          958      975      +17     
+ Misses        191      186       -5     

Impacted Files	Coverage Δ
src/access.rs	86.02% <33.33%> (-7.81%)	:arrow_down:
src/lib.rs	87.56% <0.00%> (+0.51%)	:arrow_up:
src/strategy/hybrid.rs	93.84% <0.00%> (+6.15%)	:arrow_up:
src/debt/helping.rs	89.06% <0.00%> (+10.93%)	:arrow_up:

[steffahn]

@vorner I’ve addressed the unrelated CI failures (I don’t know if that’s the "proper" way to do this with the dependencies for 1.45 compilation, but it seems like it works); feel free to re-run CI.

[steffahn]

are there any cases left where it is necessary?

as already noted over in #77:

the actual AccessConvert is somewhat more general, as it allows – say – Box<dyn Foo> for a trait Foo: DynAccess, too.

Similarly, you can use it for things like Box<dyn DynAccess + Unpin> or Box<dyn DynAccess + Sync> or other more rare variantions on the trait object type that I didn't cover with any explicit implementation. (I chose the same selection three of auto-trait combinations (none, only send, and send+sync) that e. g. dyn Any in the standard library supports for downcasting; if there are others with practical use, they can be added later.)

[vorner]

Hello again. Just wanted to check, are you still working on the docs, or did that one get lost in the rest of the discussion? 😇 Just that we both don't wait on something from the other side.

[steffahn]

Thanks for the ping. I didn’t misunderstand, but I haven’t written the docs yet either; I’ll try to allocate some time to this soon :-)

[vorner]

Hello.

I've discovered this still sits here. Not that I would have too much free time myself, but should I just take it over and write the docs, so it moves forward?

[steffahn]

I forgot this PR even exists, so ehm... sure, feel free to take over, thank you.

[vorner]

I've taken the main commit, put it onto current (which already had the CI issues mostly solved) master and made some docs for it and merged in #85.

I'll try to collect few more outstanding PRs and release a new version soon. If I don't get around to it by the end of the year, feel free to ping me.

Thank you for discovering the way to do this :-)