[Gm2] compiler bugs

[Gaius Mulley]

Hi,

ahh I've just discovered two problems with some additional test code.

Firstly that the compiler is taking too long.  Secondly that it keeps
hitting sigsegv in tree.c.  The first problem I've known about since
the type declaration rewrite in April.  The solution is to write a
straightforward dynamic set module similar to
gm2/gm2-compiler/Lists.{def,mod} - currently Lists are being
used as sets in M2GCCDeclare.  I believe this will fix the speed
problem - but profiling the compiler will prove this hypothesis.

The second bug I've seen in the regression tests - and it needs to be
fixed urgently.. I hadn't thought it would be so widely hit..

regards,
Gaius