Incorrect calculation of bitmap buffer length

[TimNN]

The length of the bitmap buffer is calculated using self.width() * self.rows(), this however assumes that one pixel equals one byte, which might not be the case (for example if pixel_mode == Mono. The correct way to calculate the buffer length would be self.pitch().abs() * self.rows(), if I understand the freetype documentation correctly.

Also, if I interpret the documentation correctly, when pitch is negative, the buffer pointer points at the last row in memory (i.e. buffer points at the row with index zero, but the row at index one is located at a lower address than that row in memory).

[zummenix]

Looks like this fixed in #164. Closing.

[Kagami]

Start of the buffer in case of negative pitch is not yet fixed.

[Kagami]

Hm, no, start of the buffer seems to be always the same. The bitmap is just stored upside-down in case of negative pitch, but last address is still at bitmap.buffer + len. @m4rw3r, correct?

[m4rw3r]

@Kagami If I read what I wrote in this pull-request correctly it should be upside down with negative pitch, yes: #164