Fix overflow parsing large float exponents

[skeeto]

Parsing huge exponents results in a signed overflow (UB). In practice it likely wraps to an incorrect KDL number for the input. The fix includes a test that fails without it.

---

To demonstrate the bug two ways at once:

$ CFLAGS=-fsanitize=undefined cmake -B build
$ cmake --build build
$ echo 'x 0e10000000000' | build/src/utils/ckdl-cat 
src/parser.c:1051:55: runtime error: signed integer overflow: 1000000000 * 10 cannot be represented in type 'int'
x 1.0

First, UBSan detects signed overflow, then an incorrect result of 1.0 due to wrapping.

[tjol]

Ok, wow. It took me a moment to understand what the actual issue was. Beautiful and astonishing UBsan find!