Update README.md

[Kapsonfire]

Possible to save size in BigInt Handling, when using Number if its in SAFE range

[gfx]

Thanks for your suggestion, but I'm not sure it's a good idea. Rather, I suspect it introduces unnecessary complexity because JavaScript's BigInt is not compatible with the primitive number. There might be cases where that trick is useful, but I believe we shouldn't do that in general.

[Kapsonfire]

Thats why it checks for the SAFE Representations. These are always safe to work with

[gfx]

I said "not compatible" because you cannot mix BigInt and Number in arithmetic operations:

const x = BigInt(1) + 1; // => TypeError: Cannot mix BigInt and other types, use explicit conversions

So if you decode a MessagePack binary in this way, you must check the type of values that might be a BigInt. On the other hand, if a particular field is always a BigInt, you don't need to check its type.

[Kapsonfire]

But its always BigInt. The Decoder always returns BigInt, even if its encoded as number and not as string.
Your argument is only valid, if the Decoding side doesnt decode with the same CodecExtension. But thats true for every CodecExtension

[gfx]

Oh, I'm sorry! I misunderstood this PR. This PR changes only how BigInt is encoded in order to optimize the encoded binary!

Okay, so it looks good to me. Could you update codec-bigint.test.ts as well?

[Kapsonfire]

Done :-)

[gfx]

Thank you for your contribution! Will merge it after CI passes.