v0.922

Notes for 0.922 release:

In the table class, the set() function no longer automatically adds
rows or columns. This prevents some behavior which was not intuitive for
users.

Several tensor functions have been added, including
tensor::swap_data(), tensor_out(), tensor::min(), tensor::max(),
tensor::copy(), tensor::is_valid(). Slicing for tensor_grid objects
has been improved. Copy constructors for tensor and tensor_grid have
been created.

A few long double integration classes have been added when higher
accuracy is required.

The MCMC classes have been completely reworked, offering better
support for OpenMP and MPI parallelism.

The acol command has been expanded and improved in several ways.
Code for o2graph has been moved out and the documentation has
been improved.

The conversion functions in convert_units now has a const version

The documentation has been improved everywhere.

A new nearest neighbor interpolation type has been added.

A seed bug in rng_gsl has been fixed.

HDF5 I/0 for table3d types now allows const references.

The cloud_file class is updated and simplified.

The prob_dens_func_amr class has been updated.

Some work has been done on heat capacity functions for particle
objects.

The fermion_rel::calc_mu() and pair_mu() functions work better
now with bad initial guesses.

The tov_solve, eos_tov, eos_cs2_const, and eos_had_rmf_hyp classes
are improved.