fix: add specialized promotion rule for SArray and MArray

[henry2004y]

Attempt to fix #144

[github-actions]

FastInterpolations.jl Benchmarks

All benchmarks (50 total, click to expand)

Benchmark	Current: 2b4d5ff	Previous	Imm. Ratio	Grad. Ratio
10_nd_construct/bicubic_2d	37470.0 ns	37996.0 ns	0.986	0.967
10_nd_construct/bilinear_2d	631.4 ns	751.5 ns	0.84	0.948
10_nd_construct/tricubic_3d	357077.0 ns	337158.0 ns	1.059	1.046
10_nd_construct/trilinear_3d	1703.6 ns	1999.8 ns	0.852	0.904
11_nd_eval/bicubic_2d_batch	1429.7 ns	1529.2 ns	0.935	0.868
11_nd_eval/bicubic_2d_scalar	16.3 ns	15.9 ns	1.025	1.01
11_nd_eval/bilinear_2d_scalar	7.3 ns	7.0 ns	1.043	1.014
11_nd_eval/tricubic_3d_batch	3243.0 ns	3457.1 ns	0.938	0.948
11_nd_eval/tricubic_3d_scalar	33.4 ns	35.8 ns	0.931	0.954
11_nd_eval/trilinear_3d_scalar	13.6 ns	13.0 ns	1.047	1.027
12_cubic_eval_gridquery/range_random	4227.3 ns	4648.9 ns	0.909	0.943
12_cubic_eval_gridquery/range_sorted	4213.5 ns	4645.5 ns	0.907	0.942
12_cubic_eval_gridquery/vec_random	9006.8 ns	10054.4 ns	0.896	0.937
12_cubic_eval_gridquery/vec_sorted	3212.8 ns	3200.1 ns	1.004	1.001
13_nd_oneshot_gridquery/bicubic_2d_rand_rand	66010.4 ns	65912.3 ns	1.001	-
13_nd_oneshot_gridquery/bicubic_2d_sort_rand	62813.5 ns	61742.1 ns	1.017	-
13_nd_oneshot_gridquery/bicubic_2d_sort_sort	58831.0 ns	58935.1 ns	0.998	-
13_nd_oneshot_gridquery/bilinear_2d_rand_rand	20191.9 ns	16653.4 ns	1.212	-
13_nd_oneshot_gridquery/bilinear_2d_sort_rand	9367.3 ns	9097.8 ns	1.03	-
13_nd_oneshot_gridquery/bilinear_2d_sort_sort	5686.2 ns	4859.0 ns	1.17	-
14_series_oneshot_batch/constant_inplace_vec_k8_q1000_rand	19180.8 ns	20604.4 ns	0.931	-
14_series_oneshot_batch/linear_inplace_vec_k8_q1000_rand	17808.3 ns	18261.0 ns	0.975	-
1_cubic_oneshot/q00001	529.8 ns	567.0 ns	0.934	1.106
1_cubic_oneshot/q10000	43546.4 ns	47867.7 ns	0.91	0.698
2_cubic_construct/g0100	1392.0 ns	1535.9 ns	0.906	0.987
2_cubic_construct/g1000	12759.9 ns	14363.2 ns	0.888	0.943
3_cubic_eval/q00001	20.2 ns	23.3 ns	0.868	0.9
3_cubic_eval/q00100	441.0 ns	483.7 ns	0.912	0.944
3_cubic_eval/q10000	42668.8 ns	47029.5 ns	0.907	0.943
4_linear_oneshot/q00001	24.7 ns	27.1 ns	0.912	0.909
4_linear_oneshot/q10000	18736.0 ns	19406.7 ns	0.965	0.979
5_linear_construct/g0100	35.4 ns	40.0 ns	0.885	0.92
5_linear_construct/g1000	266.2 ns	270.2 ns	0.985	0.99
6_linear_eval/q00001	10.1 ns	10.3 ns	0.981	0.979
6_linear_eval/q00100	195.8 ns	199.7 ns	0.98	0.984
6_linear_eval/q10000	18497.6 ns	19164.3 ns	0.965	0.979
7_cubic_range/scalar_query	8.3 ns	9.2 ns	0.902	0.939
7_cubic_vec/scalar_query	10.4 ns	10.2 ns	1.02	0.998
8_cubic_multi/construct_s001_q100	650.8 ns	680.8 ns	0.956	1.123
8_cubic_multi/construct_s010_q100	4441.1 ns	4983.4 ns	0.891	0.964
8_cubic_multi/construct_s100_q100	39701.2 ns	44607.8 ns	0.89	0.934
8_cubic_multi/eval_s001_q100	809.5 ns	856.3 ns	0.945	1.111
8_cubic_multi/eval_s010_q100	1792.9 ns	1832.7 ns	0.978	1.017
8_cubic_multi/eval_s010_q100_scalar_loop	2289.1 ns	2474.3 ns	0.925	0.951
8_cubic_multi/eval_s100_q100	11314.2 ns	11814.5 ns	0.958	0.981
8_cubic_multi/eval_s100_q100_scalar_loop	3322.2 ns	3508.1 ns	0.947	0.938
9_nd_oneshot/bicubic_2d	46215.4 ns	46253.3 ns	0.999	1.145
9_nd_oneshot/bilinear_2d	538.4 ns	587.1 ns	0.917	0.919
9_nd_oneshot/tricubic_3d	429017.9 ns	386974.7 ns	1.109	1.175
9_nd_oneshot/trilinear_3d	1035.9 ns	1073.5 ns	0.965	0.957

⚠️ Performance Regression Confirmed ⚠️

After re-running 7 flagged benchmark(s) 10 time(s), 7 regression(s) confirmed.

Benchmark	Current	Previous	Imm. Ratio	Grad. Ratio	Tier
13_nd_oneshot_gridquery/bilinear_2d_rand_rand	20191.9 ns	16653.4 ns	1.212	-	immediate
13_nd_oneshot_gridquery/bilinear_2d_sort_sort	5686.2 ns	4859.0 ns	1.17	-	immediate
1_cubic_oneshot/q00001	529.8 ns	567.0 ns	0.934	1.106	gradual
8_cubic_multi/construct_s001_q100	650.8 ns	680.8 ns	0.956	1.123	gradual
8_cubic_multi/eval_s001_q100	809.5 ns	856.3 ns	0.945	1.111	gradual
9_nd_oneshot/bicubic_2d	46215.4 ns	46253.3 ns	0.999	1.145	gradual
9_nd_oneshot/tricubic_3d	429017.9 ns	386974.7 ns	1.109	1.175	both

Thresholds: immediate > 1.1x (vs latest master), gradual > 1.1x (vs sliding window)

This comment was automatically generated by Benchmark workflow.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.45%. Comparing base (51d3afb) to head (2b4d5ff).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #145   +/-   ##
=======================================
  Coverage   96.45%   96.45%           
=======================================
  Files         143      143           
  Lines       11957    11962    +5     
=======================================
+ Hits        11533    11538    +5     
  Misses        424      424           

Files with missing lines	Coverage Δ
src/core/utils.jl	89.47% <100.00%> (+0.48%)	⬆️

🚀 New features to boost your workflow:

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

[mgyoo86]

@henry2004y Thanks for tackling this.
Closing in favor of #146 which takes a more generic trait-level approach.