Add a helper function for assembling a snia graph

[jeremykubica]

Instead of having users always manually recreate all the nodes needed for each graph, we can prebuild sections of the graph. snia_x0_x1_from_host() creates and links the nodes used by snia models.

[github-actions]

Before [1f3a6d9]	After [9c7f1eb]	Ratio	Benchmark (Parameter)
6.52±0.8ms	5.87±0.9ms	~0.90	benchmarks.TimeSuite.time_chained_evaluate
2.63±0.9ms	2.03±0.9ms	~0.77	benchmarks.TimeSuite.time_fnu_to_flam
1.53±0.09ms	1.54±0.04ms	1.01	benchmarks.TimeSuite.time_apply_passbands
5.11±0.07ms	5.17±0.06ms	1.01	benchmarks.TimeSuite.time_evaluate_salt3_passbands
11.0±0.5ms	11.0±0.1ms	1.00	benchmarks.TimeSuite.time_evaluate_salt3_model
8.72±0.03ms	8.72±0.1ms	1.00	benchmarks.TimeSuite.time_load_passbands
28.7±0.5μs	28.8±0.3μs	1.00	benchmarks.TimeSuite.time_make_new_salt3_model
1.38±0.01s	1.37±0.01s	0.99	benchmarks.TimeSuite.time_make_x1_from_hostmass
128±0.9μs	127±1μs	0.99	benchmarks.TimeSuite.time_sample_x0_from_distmod
16.6±0.3μs	16.4±0.4μs	0.99	benchmarks.TimeSuite.time_sample_x1_from_hostmass

Click here to view all benchmarks.

[mi-dai]

This looks good to me. My concern is now we lose the flexibility to define x1_func in other ways since it's already pre-defined in this snia_x0_x1_from_host function.

[jeremykubica]

It is a trade off. The question for you is whether it is worth having this as a default implementation or whether the user should specify it themselves in the simulation (as in your demo notebook). I think either way will be fine.

[mi-dai]

I think it's better to keep the way in the demo notebook, because it shows clearly how each function is defined. If user would like to keep all pre-defined functions we should wrap everything into one function instead of only wrapping x0 and x1.

[jeremykubica]

Sounds good. Closing this PR.