Add another kernel test, fix small bugs

[water111]

No description provided.

[coveralls]

Pull Request Test Coverage Report for Build 414368845

• 19 of 19 (100.0%) changed or added relevant lines in 3 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.01%) to 81.245%

---

Totals
Change from base Build 409669359:	0.01%
Covered Lines:	17795
Relevant Lines:	21903

---

💛 - Coveralls

[water111]

The windows test failures found a very real bug in the kernel implementation that the linux tests didn't catch:

(set! temp sp)   ;; store stack pointer in temp
(.sub temp off)  ;; subtract off GOAL pointer offset
;; incorrect code: 
 (set! (-> obj sp) (the int sp)) ;; store GOAL pointer to stack
;; correct code:
 (set! (-> obj sp) (the int temp)) ;; store x86-64 pointer to stack

This happened to work on Linux because (-> obj sp) is a 32-bit integer, and the location of the GOAL main memory happened to be one where you can just mask off the upper 32-bits of an x86-64 address to get the GOAL address. On windows this didn't work, because the offset wasn't all zeros in the lower 32 bits.

To make these bugs easier to find in the future, I've changed the address hint passed into mmap to one that will cause problems in these cases.

[water111]

looks like an identical bug in Debugger.cpp with the address used for breakpoints.