Add #[inline] to allow for cross-crate inlining

[cesarb]

Without #[inline], the function definition is not exported for inlining by other crates. For the byte order conversion functions, inlining can make a great difference in code quality.

For example, trying this small test crate:

extern crate byteorder;

use byteorder::{ByteOrder, LittleEndian};

pub fn example(buf: &[u8; 8]) -> u64 {
    LittleEndian::read_u64(buf)
}

The result without #[inline] in the byteorder crate:

_ZN7example20h618e367dcb99bd17iaaE:
        .cfi_startproc
        cmpq    %fs:112, %rsp
        ja      .LBB0_2
        movabsq $24, %r10
        movabsq $0, %r11
        callq   __morestack
        retq
.LBB0_2:
        subq    $24, %rsp
.Ltmp0:
        .cfi_def_cfa_offset 32
        movq    %rdi, 8(%rsp)
        movq    $8, 16(%rsp)
        leaq    8(%rsp), %rdi
        callq   _ZN22LittleEndian.ByteOrder8read_u6420hc703c516084f60e83EaE@PLT
        addq    $24, %rsp
        retq

The result after adding #[inline] to the byteorder crate:

_ZN7example20h52794dd36d3f283eiaaE:
        .cfi_startproc
        movq    (%rdi), %rax
        retq

As you can see, LittleEndian::read_u64(...) on a little-endian processor, when the compiler can determine statically that the buffer is large enough, can be inlined into a single "load" instruction.

The same should apply to all the functions I marked as #[inline] in this pull request.

[BurntSushi]

Superb. I buy it. Thanks!

[BurntSushi]

I've included this fix in 0.3.11 on crates.io. Thanks again!