Filter-functions

What's Changed
* n_coeffs_deriv sorting by thangleiter in https://github.com/qutech/filter_functions/pull/83
* Cache more intermediates in calculation with unitaries by thangleiter in https://github.com/qutech/filter_functions/pull/84
* Small improvements by thangleiter in https://github.com/qutech/filter_functions/pull/85


**Full Changelog**: https://github.com/qutech/filter_functions/compare/v1.1.1...v1.1.2

Obligatory bug fix release:

- Fix Bloch spheres in documentation (81)

- Updated readme with published paper references (70, 74)
- Added Python 3.10 support (75)
- Removed Python 3.6 support (76)
- Added colored Bloch sphere trajectories (78)

This release adds a few small features and bug fixes.

- Add `__len__` and `__getitem__` methods to `PulseSequence` (67)
- Switch CI to Github Actions (69)
- A bunch of other small improvements (66)

v1.0.3-test

* Hotfix a bug when concatenating pulses with zero `n_coeffs` (59)
* Cache on CI (60)
* Make duration and absolute times of a `PulseSequence` lazily evaluated properties (61)

Hotfixes a bug when calculating the infidelity before the infidelity derivative (58).