Client refactor

[urgentquest]

• Fix lots of TODOs by implementing client-side reply validator func validateReply().
• While I am here: add IsOk() and GetResult() to replyParser.go. both are quite frequently used in the library
• Handle edge cases with no username/password in Client.hello()
• Replace fmt.ErrorF() with errors.New() where no formatting requested
• go-staticcheck

[eyedeekay]

Yeah, simplify simplify simplify. I assume you ran all of the go tests with go test --tags nettest?

[urgentquest]

If I recall correctly I did, but let me double check this. Just to be sure

[urgentquest]

Well. There were some complications. Don't know how I missed that.
Still there is a single test failing

Initializing new ID: 2095128466
2025/05/09 00:43:36 > "HELLO VERSION MIN=3.0 MAX=3.1\n"
2025/05/09 00:43:36 < "HELLO REPLY RESULT=OK VERSION=3.1\n"
2025/05/09 00:43:36 > "NAMING LOOKUP NAME=!(@#)\n"
2025/05/09 00:43:36 < "NAMING REPLY RESULT=INVALID_KEY NAME=!(@#)\n"
--- FAIL: TestClientLookupInvalid (0.00s)
    naming_test.go:29: client.Lookup() should throw an ResultKeyNotFound error.
        Got:ReplyError: Result:INVALID_KEY - Reply:&{Topic:NAMING Type:REPLY From: To: Pairs:map[NAME:!(@#) RESULT:INVALID_KEY]}!=KEY_NOT_FOUND

Apparently i2pd I'm running says invalid key instead of result key not found which causes the failure. I could work around this by adding an extra condition. Or just leave is as it is.

[eyedeekay]

Thought about it, go ahead and add the extra condition for i2pd. I think we need to be compatible with both.

[urgentquest]

Done.
All green

[eyedeekay]

Excellent thanks again for your help/contribution. Looks good, I'm ready to merge.

At some point we might be ahead to build a SAM API test harness which runs a fake SAMv3 server that replies with accurate errors. I have one started around here in the go-i2p/go-sam-server namespace but it's never been a big priority, as we'd have to develop it, make sure it behaves exactly as the Java AND I2P SAMv3 servers, and keep up with the SAMv3 spec, but it would let us run the whole test suite in CI and catch stuff like this. Maybe the calculus on that should change soon.