[math] math::Matrix class size limitation

[hshose]

There are a few places in the matrix_impl.hpp, like here, where the matrix size is limited to 256 elements, if I see correctly. This is a 16 by 16 matrix which is very small - even on a microcontroller.

Is there a reason for having this?

Otherwise changing, e.g.,

	for (uint_fast8_t i = 0; i < getNumberOfElements(); ++i) { element[i] = data[i]; }

to

	for (uint_fast16_t i = 0; i < getNumberOfElements(); ++i) { element[i] = data[i]; }

would make sense?

If not, it would probably be good to explicitly state this limit in the documentation?

Also, maybe change the template from template<typename T, uint8_t ROWS, uint8_t COLUMNS>
to template<typename T, uint16_t ROWS, uint16_t COLUMNS> alltogether, which would probably cover all relevant matrix applications on microcontrollers? (I would see applications where one might want to use a matrix that has more than 256 entries in one dimension...).

[salkinium]

You can change it as you wish.

However, I don't want to maintain so much custom code over the long term, I would like to externalize this math code by maybe integrating eigen 3.4, which recently supports C++14. It doesn't seem to use C++ exceptions, and it mostly looks quite efficient.