[MLton] profiler bug?

[Stephen Weeks]

> Notice the large variation in total allocation between the two
> instances of the checksum benchmark.

I agree with Matthew.  One would expect this given the flags you've
used.

> If you run the same executables with @mlton gc-summary --  the numbers do
> not match up.

I do not see this.  When I run the version compiled with all the
flattening dropped, gc-summary reports

	total allocated: 45,010,011,492 bytes

while mlprof reports

	45,010,009,860 bytes allocated (1,252 bytes by GC)

On the other hand, when I run the verion compiled normally, gc-summary
reports

	total allocated: 10,010,148 bytes

while mlprof reports

	10,008,856 bytes allocated (1,252 bytes by GC)

So, there is the usual very small difference due to the runtime seeing
a little bit more allocation than the profiler, but that's all I see.