Fix panic for negative inputs to timestamp_millis.

[cmars]

This patch fixes the case where a negative millisecond offset is passed
to Timezone::timestamp_millis and timestamp_millis_opt and adds a test
case for it. Without this patch, calling timestamp_offset with a
negative value will panic with an overflow like this:

My editor (rust.vim) applies rustfmt automatically on save, which seems to have done some reformatting. Happy to back that out and reformat some other way if you prefer.

---- tests::test_parse_samples stdout ----
thread 'tests::test_parse_samples' panicked at 'attempt to multiply with
overflow',
/home/c/.cargo/registry/src/github.com-1ecc6299db9ec823/chrono-0.4.6/src/offset/mod.rs:349:34
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a
verbose backtrace.
stack backtrace:
   0: std::sys::unix::backtrace::tracing::imp::unwind_backtrace
             at libstd/sys/unix/backtrace/tracing/gcc_s.rs:49
   1: std::sys_common::backtrace::print
             at libstd/sys_common/backtrace.rs:71
             at libstd/sys_common/backtrace.rs:59
   2: std::panicking::default_hook::{{closure}}
             at libstd/panicking.rs:211
   3: std::panicking::default_hook
             at libstd/panicking.rs:221
   4: std::panicking::rust_panic_with_hook
             at libstd/panicking.rs:477
   5: std::panicking::continue_panic_fmt
             at libstd/panicking.rs:391
   6: rust_begin_unwind
             at libstd/panicking.rs:326
   7: core::panicking::panic_fmt
             at libcore/panicking.rs:77
   8: core::panicking::panic
             at libcore/panicking.rs:52
   9: chrono::offset::TimeZone::timestamp_millis_opt
             at
/home/c/.cargo/registry/src/github.com-1ecc6299db9ec823/chrono-0.4.6/src/offset/mod.rs:349
  10: chrono::offset::TimeZone::timestamp_millis
             at
/home/c/.cargo/registry/src/github.com-1ecc6299db9ec823/chrono-0.4.6/src/offset/mod.rs:327

[quodlibetor]

Awesome, thanks!