Miscellaneous fixes and additions

[chrisbrahms]

1. Correct error in CSpline for out-of-bounds evaluation
2. Add a function in Fields to optimise the CEP for maximum field strength
3. save_dict_h5 now accepts NamedTuples as well
4. Add aperture_filter which approximately calculates the transmission through an aperture for a given core size and distance from exit face
5. Add fwhm_t_allmodes statistic which sums up the field in all modes before taking the FWHM
6. Change Modes.transmission to z-integrate the loss rather than assuming it's constant--important for tapers (cc @balazsplosz )

All tests pass on my machine except for the ones mentioned in #210

[jtravs]

Looks good to me. Your ZDW change reminded me of a couple of bugs @balazsplosz found a few months back though. I will try to dig them out/remember them.

[balazsplosz]

I just rerun those simulations.

The ZDW for these parameters are not plotted. They are "off the charts"?

Parameters:

a = 1933.0863075265351e-6 # cross section [m]
gas = :HeJ # gas used
pres1 = 13.982819139614504 # pressure inside fiber [bar]
pres2 = 0.0 # pressure inside fiber [bar]
τfwhm = 30e-15 # pulse width (full width half max) [m]
λ0 = 1800e-9 # peak input wavelength [m]
energy = 1200e-6 # pulse energy [J]
L = 18.77907127064365 # length of fiber [m]

Around this range.

[jtravs]

Apologies Chris, for highjacking this PR to discuss this issue. We should have opened a new one.

If I remember correctly, the issues were:

• handling of missing in the HDF5 backend (when no ZDW found)
• making the lower bound lb in the ZDW search depedn on the grid.