Allow custom Parsers

[Jolanrensen]

Allowing users to set custom parsers would be very helpful. It could allow them things like:

class User(val name: String, val age: Int) {
    override fun toString() = "user-$name-$age"
}

DataFrame.parser.addCustomParser<User> {
    if (it.startsWith("user")) {
        val (_, name, age) = it.split("-")
        User(name, age)
    } else null
}

DataFrame.readCsv(myCustomUserData)

or we could even add an enum-specific variant. This can only be done with convert at the moment. It could allow immediate parsing of CSV -> enum for instance.

[Jolanrensen]

Additionally we should keep supporting the same arguments in ParserOptions to also allow the DF API to remain functional

[Jolanrensen]

additionally a service-based parser could help with #100 and #1147

[Jolanrensen]

From #741:

Suggested API:

class MyWrapper(val value: Int)

//either global configuration
DataFrame.parsers.add(
    stringParser { if(it.endsWith("%")) it.dropLast(1).toIntOrNull()?.let { MyWrapper(it) } else null }
)

// or provide instance here 
val df = dataFrameOf("a","b")("55%", "12%").parse(/*ParserOptions(...)*/)
df["a"].type() shouldBe typeOf<MyWrapper>()
df["b"].type() shouldBe typeOf<MyWrapper>()

Should save some effort manually selecting columns for conversion or writing a complex column selector that should somehow tell column can be parsed.

dataFrameOf("a","b")("55%", "12%").convert { 
    colsAtAnyDepth()
        .colsOf<String>()
        .filter {
            it.values().all { it.toPercentageOrNull() != null } 
        } 
}.with { it.toPercentageOrNull()!! }

Instead, all you'd need to define is (String) -> YourType? function