Replace old host tree algorithm with new algorithm in JSON reader

[karthikeyann]

Description

This PR replaced old tree algorithm in JSON reader, with experimental algorithms and removed the experimental namespace.

Changes are old tree algorithm code removal, experimental namespace removal, code of scatter_offsets moved, always call new tree algorithm.
No functional change is made in this PR. All unit tests should pass with this change.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[vuule]

what is the performance impact of the switch?

[karthikeyann]

Benchmark comparison:

./benchmarks/JSON_READER_NVBENCH --benchmark json_read_io --benchmark nested_json_gpu_parser_depth --json "nested_json_gpu_parser_depth.json"
./benchmarks/JSON_READER_OPTION_NVBENCH --benchmark json_reader --json "reader_options.json"

branch-24.12 vs this PR

json_read_io

[0] NVIDIA RTX 5880 Ada Generation

io	Ref Time	Ref Noise	Cmp Time	Cmp Noise	Diff	%Diff	Status
FILEPATH	337.697 ms	0.17%	333.087 ms	0.15%	-4610.468 us	-1.37%	FAIL
HOST_BUFFER	243.569 ms	0.49%	242.734 ms	0.32%	-834.790 us	-0.34%	FAIL
DEVICE_BUFFER	156.994 ms	0.49%	156.832 ms	0.18%	-162.656 us	-0.10%	PASS

nested_json_gpu_parser_depth

[0] NVIDIA RTX 5880 Ada Generation

depth	string_size	Ref Time	Ref Noise	Cmp Time	Cmp Noise	Diff	%Diff	Status
2^1	2^20	1.643 ms	0.42%	1.621 ms	0.69%	-21.960 us	-1.34%	FAIL
2^2	2^20	1.643 ms	0.76%	1.625 ms	0.50%	-17.678 us	-1.08%	FAIL
2^3	2^20	5.249 ms	0.33%	5.149 ms	0.67%	-100.062 us	-1.91%	FAIL
2^4	2^20	6.681 ms	0.50%	6.503 ms	0.50%	-177.942 us	-2.66%	FAIL
2^1	2^22	1.900 ms	0.68%	1.887 ms	0.66%	-12.583 us	-0.66%	PASS
2^2	2^22	1.901 ms	0.50%	1.884 ms	0.92%	-16.896 us	-0.89%	FAIL
2^3	2^22	5.635 ms	0.54%	5.464 ms	0.50%	-170.811 us	-3.03%	FAIL
2^4	2^22	7.072 ms	0.50%	6.907 ms	0.50%	-164.793 us	-2.33%	FAIL
2^1	2^24	3.149 ms	0.58%	3.134 ms	0.58%	-15.353 us	-0.49%	PASS
2^2	2^24	3.148 ms	0.49%	3.132 ms	0.50%	-15.425 us	-0.49%	PASS
2^3	2^24	6.982 ms	0.50%	6.848 ms	0.50%	-134.010 us	-1.92%	FAIL
2^4	2^24	8.746 ms	0.50%	8.577 ms	0.50%	-168.694 us	-1.93%	FAIL
2^1	2^26	12.742 ms	0.24%	12.747 ms	0.21%	5.865 us	0.05%	PASS
2^2	2^26	12.726 ms	0.25%	12.745 ms	0.23%	18.322 us	0.14%	PASS
2^3	2^26	14.093 ms	0.17%	13.922 ms	0.17%	-170.748 us	-1.21%	FAIL
2^4	2^26	17.765 ms	0.37%	17.595 ms	0.36%	-170.727 us	-0.96%	FAIL
2^1	2^28	55.229 ms	0.23%	55.410 ms	0.28%	180.783 us	0.33%	FAIL
2^2	2^28	55.290 ms	0.19%	55.228 ms	0.30%	-61.586 us	-0.11%	PASS
2^3	2^28	52.710 ms	0.50%	52.806 ms	0.60%	96.309 us	0.18%	PASS
2^4	2^28	71.135 ms	0.91%	71.166 ms	0.89%	31.242 us	0.04%	PASS
2^1	2^30	224.089 ms	0.14%	224.347 ms	0.17%	257.710 us	0.12%	PASS
2^2	2^30	224.063 ms	0.16%	224.314 ms	0.11%	251.106 us	0.11%	FAIL
2^3	2^30	225.992 ms	0.26%	226.354 ms	0.22%	362.578 us	0.16%	PASS
2^4	2^30	302.261 ms	0.64%	302.221 ms	0.70%	-40.352 us	-0.01%	PASS

json_reader

[0] NVIDIA RTX 5880 Ada Generation

json_lines	Ref Time	Ref Noise	Cmp Time	Cmp Noise	Diff	%Diff	Status
YES	236.325 ms	0.21%	236.223 ms	0.39%	-101.778 us	-0.04%	PASS
NO	236.867 ms	0.41%	235.902 ms	0.31%	-964.772 us	-0.41%	FAIL

@vuule Performance impact is minimal. Slightly faster on smaller size jsons.

[karthikeyann]

/merge