std.fmt: add option to explicitly show positive number signs; show sign before leading zeroes

[HydroH]

Closes #14436 .
This PR implemented following features:

• Added an option to explicitly show + sign when formatting positive numbers;
• When the fill character is 0 and number aligned to right, they will be recognized as leading zeroes, and the sign will be appended before the leading zeroes;
• For hexadecimal floating point numbers, leading zeroes will be inserted after 0x.
  e.g.

try std.testing.expectFmt("  +1", "{:+>4}", .{1});
try std.testing.expectFmt("+001", "{:+0>4}", .{1});
try std.testing.expectFmt("+0x001p0", "{x:+0>8}", .{1.0});

Points worth discussing:

• How to specify + as the fill character when the signed option is not present? Currently I use \\+ as the indentifier for such.
• What should be the behaviour when formating inf and nan with leading zeroes and a sign? Currently the sign is still placed before leading zeroes.
  e.g.

try std.testing.expectFmt("1+++", "{:\\+<4}", .{1});
try std.testing.expectFmt("+00nan", "{:+0>6}", .{std.math.nan(f64)});

[xdBronch]

what about putting the+ at the end to avoid the \\+? its a little messy and not straightforward when it actually is used for fill. e.g. {:0>4+}

[castholm]

This is just my opinion, but special-casing right-aligned zero fills feels like a bit of a hack and it's conflating generic alignment and numeric formatting which are two very different things. A user might want to zero-pad a signed number and still align that number separately, e.g.

// zero-pad integer part to 3 digits, print positive sign, left-align and fill with spaces to a min width of 10

|-000.100  |
|+010.333  |
|-097.167  |

It would be a major breaking change and probably worth discussing in a proposal issue before putting work in, but I think we should consider whether format specifiers should be updated to distinguish between generic alignment options and numeric formatting options. Perhaps something like {[argument][specifier]:[fill][alignment][width]#[+][min_integers].[precision]} could be used as starting point. For the example above the format string would then be {d: <10#+3.3}.

[andrewrk]

I don't know what to do with this. Perhaps we can reach a consensus over at #19488 and then decide.

[andrewrk]

Sorry, it bitrotted before I got around to it.