Add write_json to improve compatibility

[nhz2]

This PR adds a JSON.write_json function with the same functionality and keyword arguments as the new JSON.json function. This function is provided for backward compatibility when upgrading from
older versions of JSON.jl where the json function signature differed.

The idea is that JSON.write_json will be backported to v0.21.5

EDIT: Changed the name from writefile to write_json

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.20%. Comparing base (ac46743) to head (f25c33a).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #381      +/-   ##
==========================================
+ Coverage   90.10%   90.20%   +0.10%     
==========================================
  Files           7        7              
  Lines        1334     1348      +14     
==========================================
+ Hits         1202     1216      +14     
  Misses        132      132              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[quinnj]

I'm not following why we want this or the 2nd definitions? writefile that doesn't write to a file and just returns a byte vector?

[nhz2]

I want to avoid extra conversions between String and Vector{UInt8} for use with Zarr.jl and SmallZarrGroups.jl

I'm also trying to match the style of the writefile function in Parquet2.jl https://expandingman.gitlab.io/Parquet2.jl/#Writing-Data

[nhz2]

Yes, looking at this again, writefile isn't a good name for the function, so I changed it to write_json.

[nhz2]

Here is a benchmark on the cost of doing the extra String to Vector{UInt8} conversion:

julia> using Chairmarks, JSON, Random

julia> @be randstring(10000) Vector{UInt8}(JSON.json(_)) seconds=2
Benchmark: 52143 samples with 4 evaluations
 min    4.807 μs (7 allocs: 19.852 KiB)
 median 4.874 μs (7 allocs: 19.852 KiB)
 mean   5.566 μs (7 allocs: 19.856 KiB, 0.82% gc time)
 max    15.658 ms (7 allocs: 19.883 KiB, 99.92% gc time)

julia> @be randstring(10000) JSON.write_json(Vector{UInt8}, _) seconds=2
Benchmark: 50390 samples with 7 evaluations
 min    2.971 μs (4 allocs: 9.914 KiB)
 median 3.006 μs (4 allocs: 9.914 KiB)
 mean   3.394 μs (4 allocs: 9.919 KiB, 0.82% gc time)
 max    8.692 ms (4 allocs: 9.945 KiB, 99.92% gc time)