add book_load warnings, arbitrary-size strings/bytes

[enricozb]

• Adds support for arbitrarily-sized strings and bytes, which lets READ and WRITE calls work for any size of bytes.
• book_load now warns if definitions are too large for the respective runtime.
• removes unused tm parameter from global allocation functions for cuda (does this impact runtime b/c of variable stack sizes)?

Future Work

The rust parser should warn about definition sizes before handing it off to the runtimes.

[HigherOrderBot]

Perf run for b44023f:

compiled
========

file            runtime         main            (local)       
==============================================================
sort_bitonic    c                        3.96s           3.76s
                cuda                     0.24s           0.14s
--------------------------------------------------------------
sum_rec         c                        1.61s           1.73s
                cuda                     0.14s           0.06s
--------------------------------------------------------------
sum_tree        c                        0.12s           0.12s
                cuda                     0.10s           0.02s
--------------------------------------------------------------
tuples          c                        4.05s           3.31s
                cuda                   timeout         timeout
--------------------------------------------------------------

interpreted
===========

file            runtime         main            (local)       
==============================================================
sort_bitonic    c                        3.79s           3.83s
                cuda                     0.24s           0.15s
                rust                   timeout         timeout
--------------------------------------------------------------
sum_rec         c                        1.78s           1.75s
                cuda                     0.14s           0.06s
                rust                    13.43s          13.48s
--------------------------------------------------------------
sum_tree        c                        0.26s           0.21s
                cuda                     0.09s           0.02s
                rust                     0.83s           0.84s
--------------------------------------------------------------
tuples          c                        2.08s           3.52s
                cuda                   timeout         timeout
                rust                     3.79s           3.80s
--------------------------------------------------------------