Add CLI for migration

[JOOHOJANG]

What this PR does / why we need it:
Define semi-auto migration architecture and add migration scripts for add version vector.

Process

1. make new version directory in migrations
2. add migration script files.
3. add version into migrationMap in /cmd/migration.go

Usage example

$ ./yorkie migration --from v0.5.1 --to v0.6.0 --mongo-connection-uri mongodb://localhost:27017 --batch-size 10000 --database-name yorkie-meta

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Does this PR introduce a user-facing change?:

Additional documentation:

Checklist:

• Added relevant tests or not required
• Didn't break anything

[hackerwins]

Thanks for your contribution. Overall looks good. I have a few questions.

A. migrations directory

Currently, Yorkie is built and distributed as a single binary(yorkie). In this change, it seems that Yorkie CLI is designed to execute by referencing the migrations directory.

While migrations directory is included in the binary during the build, it is unclear why the command(yorkie migration) still needs to reference the directory.

It would be better to retrieve and execute the target scripts based on the migrationMap rather than external directory. I'm curious about your opinion.

B. Batch Processing for Find

Please, consider implementing batch processing for Find. The current approach of loading the entire collection into memory could lead to excessive memory usage.

collection := db.Database(databaseName).Collection("changes")

cursor, err := collection.Find(ctx, bson.M{})
if err != nil {
  return err
}

[JOOHOJANG]

Thanks for your contribution. Overall looks good. I have a few questions.

A. migrations directory

Currently, Yorkie is built and distributed as a single binary(yorkie). In this change, it seems that Yorkie CLI is designed to execute by referencing the migrations directory.

While migrations directory is included in the binary during the build, it is unclear why the command(yorkie migration) still needs to reference the directory.

It would be better to retrieve and execute the target scripts based on the migrationMap rather than external directory. I'm curious about your opinion.

B. Batch Processing for Find

Please, consider implementing batch processing for Find. The current approach of loading the entire collection into memory could lead to excessive memory usage.

collection := db.Database(databaseName).Collection("changes")

cursor, err := collection.Find(ctx, bson.M{})
if err != nil {
  return err
}

A. Agree, you're right. Watching migration map is enough. I'll fix it.
B. Cursor has default batch size (101 documents) so �we should not increase it unless we need a larger batch size. https://www.mongodb.com/docs/manual/tutorial/iterate-a-cursor/

[hackerwins]

Thanks for your contribution.