With this patch parse nodes are allocated sequentially in chunks.  This
reduces fragmentation of the heap and prevents waste at the end of
individually allocated parse nodes.

Saves roughly 20% of RAM during parse stage.

Python semantics are that rhs of shift must be non-negative, so there's
no need to handle negative values in the underlying mpz implementation.

This patch adds more fine grained error message control for errors when
parsing integers (now has terse, normal and detailed).  When detailed is
enabled, the error now escapes bytes when printing them so they can be
more easily seen.

To avoid local ref table overflow.

Those two are rarely used features and better to have the extra heap.

It's relatively small (between 44 and 56 bytes) and helps to reduce heap
pressure and fragmentation during compilation.

Adding a line-end makes the determination of skip_native fail as it compares
the output against b'CRASH' while it is in fact b'CRASH\n'

Saves a few bytes of code space and eliminates need for rot_two
bytecode (hence saving RAM and execution time, by a tiny bit).