ENH: Add counting triangles example

[mtsokol]

Hi @hameerabbasi,

Here's a short example that runs counting triangles with Finch, SciPy, and NetworkX. Two comments:

• The Finch implementation is the slowest right now, it's only run in eager mode as lazy mode is blocked by Matmul and reduce in lazy mode fails with ArgumentError willow-ahrens/Finch.jl#615
• I see that examples in examples/ directory aren't executed in the CI job anymore, is it intended? I think they should still be executed.

[codspeed-hq]

CodSpeed Performance Report

Merging #781 will degrade performances by 25.26%

_{Comparing triangles-example (733f44b) with main (a60f0ad)}

Summary

❌ 7 regressions
✅ 333 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	triangles-example	Change
❌	test_index_slice[side=100-rank=2-format='gcxs']	2.4 ms	3.2 ms	-25.26%
❌	test_elemwise[f=<built-in function add>-backend='Finch'-side=1000]	1.5 ms	1.7 ms	-11.83%
❌	test_elemwise[f=<built-in function add>-backend='Finch'-side=100]	728.4 µs	925.2 µs	-21.27%
❌	test_elemwise[f=<built-in function add>-backend='Finch'-side=500]	893.8 µs	1,094 µs	-18.3%
❌	test_elemwise[f=<built-in function mul>-backend='Finch'-side=1000]	750.1 µs	958.8 µs	-21.76%
❌	test_elemwise[f=<built-in function mul>-backend='Finch'-side=100]	706.6 µs	908.3 µs	-22.21%
❌	test_elemwise[f=<built-in function mul>-backend='Finch'-side=500]	723.3 µs	935.9 µs	-22.72%

[hameerabbasi]

I see that examples in examples/ directory aren't executed in the CI job anymore, is it intended? I think they should still be executed.

I believe that benchmarks should be moved to CodSpeed benchmarks, see #772, but examples with small datasets should be executed without measuring timings.

[mtsokol]

Ok, then I guess this PR is ready anyway.