use parse for BigFloat/BigInt

[simonbyrne]

This deprecates BigInt/BigFloat(::AbstractString) methods, in favour of parse (see #10594 and #10599 ).

• It also defines big_str and bigfloat_str macros (analogous to the bigint_str macro): on that note, would these be better as BigFloat_str and BigInt_str?
• I changed the default "base" argument in parse(BigFloat,) to 0: this is basically the same as 10, but allows for use of 0x/0b prefixes for hex-float/bit-float (RFC: bit-float literal #9371) formats.

[dpsanders]

Nice. Could we also extend the big function so we can write

big("0.1")

?

I am constantly writing this...

[tkelman]

should these be

@deprecate BigFloat(s::AbstractString) parse(BigFloat,s)
@deprecate BigInt(s::AbstractString) parse(BigInt,s)

?

[simonbyrne]

Good catch, thanks!

[simonbyrne]

I've made it a string macro, so you can write

big"0.1"

Is that sufficient, or would a function be useful as well?

[simonbyrne]

Okay, I've replaced bigint_str with big_str.

[pao: bit_str is thankfully not what you meant...that would have been confusing]

[StefanKarpinski]

LGTM. @JeffBezanson?

[mschauer]

Small inconsistency: tryparse() allows a base argument, but parse not.

[simonbyrne]

Ah, good point: I was just exposing the GMP/MPFR functionality. Our parse & tryparsefunctions for other types don't support different bases either.

[mschauer]

For some Int they do,
julia> parse(Int,"1A",16) 26

[simonbyrne]

Ah, good point. We should expose this then.