Split instant() to two simpler functions instant() and instantFromText()

[AloisMahdal]

Current signature of zeit.instant() is:

fn instant(io: std.Io, config: Instant.Config) !Instant;

which forces unnecessary complexity at call site, because:

• it's fallible (!Instant), but the fallibility is only applicable when parsing text, ie. when config.source is initialized with certain tags,

• and it always accepts std.Io but this is only relevant when config.source is initialized with .now tag.

This commit splits current .instant() into two simpler functions:

fn instant(source: Instant.Source, tz: *const TimeZone) Instant;

where accepted union contains std.Io only when necessary, ie. as payload for the .now tag, and

fn instantFromText(format: Instant.TextFormat, text: []const u8, tz: *const TimeZone) !Instant;

where TextFormat enum contains only recognized formats (eg. .iso8601).

Notice that in the first case, even though .instant() can talk to std.Io, the call is to system clock, which is infallible.

Also I've "smuggled in" further simplification: timezone information is passed as separate argument, instead of Instant.Config, leading to more readable call sites in both cases, as demonstrated in README.md and tests:

const now = try zeit.instant(io, .{});

⬇️

const now = zeit.instant(.{.now = io}, &zeit.utc);

..

const epoch = try zeit.instant(std.testing.io, .{ .source = .{ .unix_timestamp = 0 } });

⬇️

const epoch = zeit.instant(.{ .unix_timestamp = 0 }, &zeit.utc);

and

_ = try zeit.instant(std.testing.io, .{
    .source = .{
        .rfc3339 = "2024-03-16T08:38:29.496706064-1200",
    },
});

⬇️

_ = try zeit.instantFromText(
    .rfc3339,
    "2024-03-16T08:38:29.496706064-1200",
    &zeit.utc,
);

The readability comes at a small price that timezone, which would default to UTC in the older API, now must be passed, but this is trivial, easy to learn from the README.md and arguably better as in "more explicit".

[AloisMahdal]

Note that while looking for call sites and docs to update, I found some under the .latch sub-directory but I have no idea what it is and it seemed irrelevant so I did not touch it.

[AloisMahdal]

(force-pushing to remove period prefixes from the commit message---added by me out of weird habit.)