[ADD] Change (transposed) Jacobian to purely PyTorch

[f-dangel]

Review after #142, #145, #146.

This introduces an attribute FIXED_DATA_ORDER, which will make sure that the data loader is deterministic when set to True (which is the case for the Jacobians). This will address #75 in a future PR.

[coveralls]

Pull Request Test Coverage Report for Build 12053712305

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 106 of 112 (94.64%) changed or added relevant lines in 3 files are covered.
• 57 unchanged lines in 3 files lost coverage.
• Overall coverage decreased (-2.4%) to 89.508%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
curvlinops/_torch_base.py	68	74	91.89%

Files with Coverage Reduction	New Missed Lines	%
curvlinops/_base.py	1	83.69%
curvlinops/examples/utils.py	10	31.25%
curvlinops/fisher.py	46	26.47%

Totals
Change from base Build 11944028535:	-2.4%
Covered Lines:	1527
Relevant Lines:	1706

---

💛 - Coveralls

[runame]

LGTM!