Use fromUtf8 in web3.toHex

[wjmelements]

Fixes: #1179

This fixes an issue I am having where if I submit Korean bytes parameters through web3.js it sends garbage. It likely fixes other utf8 issues as well.

Before:

web3.toHex('내가 제일 잘 나가')
"0xb0b4ac0020c81cc77c20c79820b098ac00"
web3.toUtf8(web3.toHex('내가 제일 잘 나가'))
inpage.js:14308 Uncaught Error: Invalid UTF-8 detected
    at c (inpage.js:14308)
    at Object.decode (inpage.js:14308)
    at Proxy.toUtf8 (inpage.js:14308)
    at <anonymous>:1:6

After:

web3.toHex('내가 제일 잘 나가')
"0xeb82b4eab08020eca09cec9dbc20ec9e9820eb8298eab080"
web3.toUtf8(web3.toHex('내가 제일 잘 나가'))
"내가 제일 잘 나가"

[wjmelements]

This breaks one test case:

{ value: '\u0003\u0000\u0000\u00005èÆÕL]\u0012|<9d>ξ<9e>\u001a7«<9b>\u00052\u0011(Ð<97>Y\n<\u0010\u0000\u0000\u0000\u0000\u0000\u0000e!ßd/ñõì\f:z¦Î¦±ç·÷Í¢Ëß\u00076*<85>\b<8e><97>ñ<9e>ùC1ÉUÀé2\u001aÓ<86>B<8c>',
      expected: '0x0300000035e8c6d54c5d127c9dcebe9e1a37ab9b05321128d097590a3c100000000000006521df642ff1f5ec0c3a7aa6cea6b1e7b7f7cda2cbdf07362a85088e97f19ef94331c955c0e9321ad386428c'}

[wjmelements]

With ascii, the encoding is 0x0300000035e8c6d54c5d127c9dcebe9e1a37ab9b05321128d097590a3c100000000000006521df642ff1f5ec0c3a7aa6cea6b1e7b7f7cda2cbdf07362a85088e97f19ef94331c955c0e9321ad386428c.

I contend that 0x0300000035c3a8c386c3954c5d127cc29dc38ec2bec29e1a37c2abc29b05321128c390c297590a3c100000000000006521c39f642fc3b1c3b5c3ac0c3a7ac2a6c38ec2a6c2b1c3a7c2b7c3b7c38dc2a2c38bc39f07362ac28508c28ec297c3b1c29ec3b94331c38955c380c3a9321ac393c28642c28c is the desired utf8 encoding of the data and that the test case is incorrect.

This differs from the current behavior of fromUtf8 because code point 0 terminates the string. The test cases expect this to remain the behavior. I would expect code point 0 to not terminate the string, and for it to be represented by 00.

[coveralls]

Coverage increased (+0.01%) to 90.736% when pulling b41cdd6 on wjmelements:utf8 into 3c86456 on ethereum:develop.

[wjmelements]

I'm preserving the trimming behavior in fromUtf8 by adding an optional parameter, allowZero.

One advantage of not preserving this trimming behavior would be that we could assert utils.toUtf8(utils.fromUtf8(test.value)) == test.value in the test cases.

One reason to preserve the trimming behavior is that some conceivable client might depend on it.

[wjmelements]

Until this is determined, as a workaround you can do

contract.bytesFn(web3.fromUtf8(utf8str), (err,result)=>{...});

[frozeman]

I compared it to how i handle that in web3.js 1.0 branch and your change seems to be correct.
Thanks.