Made some tweaks to the low-level code section to enable faster db lo…

[derekn97]

1. The previous Slice function constructed a new Binary Format.Record and effectively processed the preceding portion of the binary every time it extracted a range.

• Update: Replaced it with Binary.Range to return a subset starting at an offset, so that it avoids repeatedly constructing the record-based parser.

2. The code previously converted every one-, two-, and four-byte slice to a list and then accumulated its contents. This is especially expensive for varints because it happens multiple times per row.

• Update: Defined reusable binary readers once (UInt16BEFormat, UInt32BEFormat, Float64BEFormat, ReadByte) to avoid creating a binary, converting it into a one-item list, and accumulating the list for every byte in every varint.