Fix raise_non_differentiable_error type

[ValerianRey]

By contravariance, the type of the _ parameter must be a supertype of the type of the expected parameter for a backward hook, which is tuple[Tensor, ...] | Tensor. Indeed, this hook is supposed to work even if a single Tensor is passed to it.

This was highlighted by MyPy.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Files with missing lines	Coverage Δ
...c/torchjd/aggregation/_utils/non_differentiable.py	100.00% <100.00%> (ø)

🚀 New features to boost your workflow:

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