move from lexical-core to depend on lexical-parse-float directly

[decathorpe]

Closes #49.

Code changes are minimal - I have adjusted the use statement to import things under the same names as previously, and only one line of actual code needs to be changed - the parse_partial_with_options function is from lexical-core, but it is a trivial one-line function, so it can just be replaced with calling the trait method on f64 directly.

[codecov]

Codecov Report

Merging #57 (789123c) into main (29b5833) will not change coverage.
The diff coverage is 100.00%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #57   +/-   ##
=======================================
  Coverage   95.35%   95.35%           
=======================================
  Files           8        8           
  Lines        1097     1097           
=======================================
  Hits         1046     1046           
  Misses         51       51           

Files	Coverage Δ
src/number_decoder.rs	90.52% <100.00%> (ø)

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 29b5833...789123c. Read the comment docs.

[codspeed-hq]

CodSpeed Performance Report

Merging #57 will not alter performance

_{Comparing decathorpe:main (789123c) with main (29b5833)}

Summary

✅ 54 untouched benchmarks

[samuelcolvin]

LGTM, I'll let @davidhewitt review and approve.

[decathorpe]

Sorry about the formatting, I forgot to run rustfmt after my changes.

If I read the benchmark results correctly, this small change actually speeds things up measurably? That's interesting ... well, it's one fewer function call, so the code can probably be more optimized.

[samuelcolvin]

I'm not sure if we have good float benchmarks? we do have some.

[davidhewitt]

LGTM, thanks!

[davidhewitt]

To summarise my opinion here: I think at some point we might want to remove the lexical dependency entirely, but this is good enough for downstream packaging for now. So we can postpone the bigger work until we have the resources to also maintain performance.