Correction of reading .npy files

[corentin-prigent]

Input files in .npy format are now correctly read.

In previous versions, it was considered that the array size specified in the header of the .npy file corresponded to the number of vertices in the mesh. However, since references (array entries) are given to hexahedra, this led to leaving unused every entry in the array corresponding to the last index in each dimension.
Hence, the number of vertices and hexahedra in every dimension have been increased by one.

Data reading has been modified from int16_t type to uint32_t.

This modification is not enough to ensure proper reading for every type of data entry (int64 for instance). Another update will shortly be made in order to correct this.

Some int have been changed to MMG5_int.

Added some standard output lines.

Example

A example is provided here. The file small-example.npy is 4 x 4 x 4 array containing integers from 1 to 64. Running hex2tet without the update leads to an incorrect output file: most references do not appear in the mesh. This problem was caused both by the offset between the number of hexahedra and vertices and the data type that was read.
small-example.tar.gz

Warning

Since it is necessary for mmg to have a cell-centered distribution of the references, this update renders the use of a vertex-centered level-set function unusable with the output tetrahedral mesh.

[Algiane]

Thanks