Implementing join functions

[zhezhaozz]

I use this draft PR for self-learning purposes. Do NOT merge this PR.

Implementing join functions, starting with left_join(df1, df2, by). The goal is listed by the following:

• [ ] if by is passed nothing, the function will find common columns to join
• [x] by can be passed with a single string indicating variable name
• [x] by can be passed with a vector of strings indication multi-columns joining
• [x] similar to dplyr in R, by can take expressions like (id == pid) to match columns with different names
• [x] Wrap the function into a macro

[zhezhaozz]

left_join now can be passed with expression to by argument. For example:

left_join(df1, df4, by=:(id = zid))
left_join(df1, df5, by=:([id = zid, pid = fid]))

[zhezhaozz]

@left_join can take the following arguments

@left_join(df1, df3, "id")
@left_join(df1, df2, ["id", "pid"])
@left_join(df1, df4, id == zid)
@left_join(df1, df5, [id == zid, pid == fid])

The parse_join_by is the function to parse the expression from by argument, without any string manipulation.

Next steps:

1. Allow by to be Nothing;
2. Add a parse_join parsing engine to handle more DataFrames.leftjoin arguments;
3. Add test cases

[kdpsingh]

Implemented a different way of doing this in PR #30.