Tweaks to Trivia::findNodeWhereRangeFitsIn

[Numpsy]

refs #3164 (comment)

• Use Array.tryPick rather than Array.choose |> Array.tryHead
• Use Option.orElseWith rather than 'Some betterChild -> Some betterChild'

Rather trivial in the grand scheme of things, but results of the Benchmarks project:

Before:

| Method | Mean     | Error   | StdDev   | Median   | Rank | Gen0     | Allocated |
|------- |---------:|--------:|---------:|---------:|-----:|---------:|----------:|
| Format | 174.5 ms | 8.05 ms | 23.22 ms | 164.4 ms |    1 | 333.3333 | 155.61 MB |

After:

| Method | Mean     | Error   | StdDev  | Rank | Gen0     | Allocated |
|------- |---------:|--------:|--------:|-----:|---------:|----------:|
| Format | 163.6 ms | 3.27 ms | 7.38 ms |    1 | 500.0000 | 155.17 MB |

My laptop is a bit variable with the runtime, but there' no real time change, just a small memory change (155.61MB to 155.24MB with the tryPick change, and to 155.17MB from orElseWith

[nojaf]

Do you think that would perform better? Or this is more of a "I would have written it that way" thing?

[Numpsy]

The Some betterChild -> Some betterChild appears to be allocatiing lots of extra instances of Option, which orElseWith avoids.
Maybe just doing Some _ -> betterChildNode would work as well

[nojaf]

appears to be allocatiing

Why are you getting that from? The before and after seem rather close?

[Numpsy]

The Visual Studio allocation profiler shows this with the match:

and this with orElseWith

And BDN itself say it reduces allocations from 155.24MB to 155.17MB.
Admittedly this is only a small change in memory given the number of allocations

[nojaf]

Alright, thanks for sharing!

[nojaf]

This is the real improvement, is it not?

[nojaf]

Thank you!

[nojaf]

Would you like a new release for this?
If so, please update the changelog.
Otherwise, we can leave this as is.

[Numpsy]

I don't think there's a need for a release just for this