Fix test tolerances for float16 precision in math tests

[antonwolfy]

This PR fixes test failures when testing with DPNP_TEST_ALL_INT_TYPES=1 against conda-forge's NumPy, where float16 precision is used in various scenarios requiring relaxed tolerances.

Issue

Tests were failing with tolerance violations when using int8/uint8 dtypes or when outputs have float16 precision. Float16 has approximately 3 decimal digits of precision, making the original strict tolerances (atol=1e-5, rtol=1e-7) inappropriate.

Changes

1. test_explog.py (commit f00eda8)

• Updated _rtol_dict to handle float16 output dtype
• Applied relaxed rtol=1e-3 for float16, rtol=1e-7 for other dtypes
• Affects tests: exp, expm1, exp2, log, log10, log2, log1p, etc.

2. test_hyperbolic.py (commit dbb7199)

• Updated _rtol_dict to handle float16 output dtype
• Applied relaxed rtol=1e-3 for float16, rtol=1e-6 for other dtypes
• Affects tests: sinh, cosh, tanh, arcsinh, arctanh, etc.

3. test_misc.py + testing framework (commit 9b108ec)

Framework enhancement:

• Extended _resolve_tolerance() in testing/_loops.py to support callable tolerance functions
• Callables receive signature: (result_dtype, **test_kwargs) enabling dynamic tolerance based on test parameters
• Updated documentation and validation in _check_tolerance_keys()

Test implementation:

• Added _interp_atol() method that dynamically adjusts tolerance based on dtype_x
• Returns atol=1e-3 when dtype_x is int8/uint8/float16 (float16 intermediate precision in sin())
• Returns atol=1e-5 for high-precision dtypes
• Applied to all 11 test_interp* methods

This approach was necessary because test_interp has float16 precision in intermediate computation (sin(fx)) rather than output, requiring inspection of input dtypes rather than output dtypes.

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/pull/2828/index.html

[github-actions]

Array API standard conformance tests for dpnp=0.20.0dev4=py313h509198e_21 ran successfully.
Passed: 1356
Failed: 1
Skipped: 16

[coveralls]

coverage: 81.977%. remained the same
when pulling 2013d47 on fix-tolerance-when-testing-against-cf-numpy
into 5e5dc24 on master.