(feat): add customization API for IntegrationGrid

[ilan-gold]

Description

Focused on customizing the integration domain, this PR allows the user to turn off checks for the domain and also gives a bit more freedom to the grid_func to not have to deal only with a and b in [a,b] but rather just takes is [a,b].

Summary of changes

• add disable_integration_domain_check argument where applicable
• change usage of a and b to just integration_domain

How Has This Been Tested?

• The existing suite of tests should suffice by not breaking since there is not new feature really here.

[github-actions]

Coverage Report

File	Stmts	Miss	Cover	Missing
boole_test.py	52	4	92%	128–131
gauss_test.py	66	4	94%	157–160
gradient_test.py	97	4	96%	109, 278–280
helper_functions.py	44	1	98%	314
integration_grid_test.py	69	4	94%	156–159
integration_test_functions.py	91	2	98%	50, 137
integrator_types_test.py	57	7	88%	114–120
monte_carlo_test.py	66	4	94%	162–165
rng_test.py	38	8	79%	64–71
simpson_test.py	65	4	94%	159–162
trapezoid_test.py	66	4	94%	156–159
utils_integration_test.py	107	8	93%	38–39, 200–206
vegas_map_test.py	73	5	93%	18, 155–158
vegas_stratification_test.py	50	4	92%	93–96
vegas_test.py	102	13	87%	227–241
torchquad
__init__.py	22	22	0%	1–48
torchquad/integration
base_integrator.py	49	7	86%	28, 68–71, 75, 108, 112, 117
boole.py	30	3	90%	54, 75–79
gaussian.py	54	3	94%	72, 98, 116
grid_integrator.py	108	7	94%	40, 160, 247–248, 250–251, 278
integration_grid.py	46	4	91%	47–48, 110, 113
monte_carlo.py	97	4	96%	100, 221–222, 244
rng.py	62	24	61%	72–75, 92–120, 140
simpson.py	29	3	90%	52, 71–75
utils.py	100	8	92%	100, 139, 170, 174, 181, 193, 195, 210
vegas.py	161	4	98%	95, 99–100, 188
vegas_map.py	98	6	94%	263–273
torchquad/utils
enable_cuda.py	12	6	50%	16–21
set_precision.py	24	3	88%	35–38, 68
set_up_backend.py	19	1	95%	12
TOTAL	2037	181	91%

Tests	Skipped	Failures	Errors	Time
52	0 💤	0 ❌	0 🔥	2m 22s ⏱️

[gomezzz]

Description

Focused on customizing the integration domain, this PR allows the user to turn off checks for the domain and also gives a bit more freedom to the grid_func to not have to deal only with a and b in [a,b] but rather just takes is [a,b].

Summary of changes

* add `disable_integration_domain_check` argument where applicable

* change usage of `a` and `b` to just `integration_domain`

How Has This Been Tested?

* The existing suite of tests should suffice by not breaking since there is not new feature really here.

@ilan-gold So, to make sure I remember correctly: This allows integrating multiple integrands on different domains by transforming each integral into [0,1], right?

If so:

• Does it work for the other integrators (Newton Cotes, MC)? Should it?
• Can we add a tiny example somewhere how this can be used? I don't know how to test it 🥲
• Should we maybe change the domain in one of the test cases to use this, otherwise it isn't really tested, is it?

[gomezzz]

Some smaller comments

[gomezzz]

only a minor comment, looks good :) 👍