to_csv is lossy for last ulp

```
x=0.33368629901358837

import pandas as pd
pd.DataFrame([x]).to_csv("frame.csv", index=False, header=False)
with open("frame.csv", "rt") as f:
b = f.read()
assert b == "0.33368629901358837\n", b

import datatable as dt
dt.Frame([x]).to_csv("table.csv")
with open("table.csv", "rt") as f:
b = f.read()
assert b == "C0\n0.33368629901358837\n", b <== fails
```