Add proof of concept for serializers without bytecode

[chmp]

Will implement #118

Todo:

• Date64
• Timestamp
• Dictionary<_, Utf8>
• Decimal128
• Float16
• Union
• LargeList
• List
• Map
• Null
• Struct
• Boolean
• Int8
• Int16
• Int32
• Int64
• UInt8
• UInt16
• UInt32
• UInt64
• Float32
• Float64
• Utf8
• LargeUtf8

[Ten0]

Great work, code is very readable and seems well written 😊
Trying to anticipate on the code flow of when I will plug, I did read a bit of this PR and that prompted some thoughts that may be either small potential future improvements or questions, so I figured I might as well post them here.
As far as I'm concerned these are all pretty minor or even opinionated apart from the map serialization lookups that can be optimized from copying a string then looking up in a map to a single string comparison with no string copy.
Please let me know if anything is unclear 😇

[chmp]

Great work, code is very readable and seems well written 😊 Trying to anticipate on the code flow of when I will plug, I did read a bit of this PR and that prompted some thoughts that may be either small potential future improvements or questions, so I figured I might as well post them here. As far as I'm concerned these are all pretty minor or even opinionated apart from the map serialization lookups that can be optimized from copying a string then looking up in a map to a single string comparison with no string copy. Please let me know if anything is unclear 😇

Thanks for all your suggestions. I will create an issue for them. TBH. I am currently preoccupied with other things. So not sure how soon I will get around to implement them.