fix int to float conversions

[geseq]

fixes #101
fixes #99

Due to #59 integer floats are converted to int, causing float check issues. One way to fix is to move float to int conversions to Int() accessor. This has the side effect of also fixing loss of precision. It also makes the implementation consistent with go's json unmarshal

Summary

Checklist

[Place an '[x]' (no spaces) in all applicable fields. Please remove unrelated fields.]

• Tests are passing: task test
• Code style is correct: task lint

[geseq]

looks like this'll require #107 to be merged to fix the CI failure

[hanzei]

Makes sense 👍