math/big: (*Float).Text and (*Float).Format don't support 'g' with negative precision

[griesemer]

In analogy to the strconv.FormatFloat and fmt.Printf functions using the 'g' format, (_Float).Text and (_Float).Format should support the 'g' format with a negative precision (== smallest number of digits necessary to identify the floating-point value uniquely).

The implementation is in (*Float).Text.

Once fixed, the work-around (to avoid a panic) in (*Float).Format can be removed.

[gopherbot]

CL https://golang.org/cl/14850 mentions this issue.