Keep backend running when exceptions occur

[tracyhenry]

Previously, the backend will throw up whenever there is an exception. The user has to kill docker containers, restart and face possibly corrupted database stuff. This issue has been tracked in #15

This PR addresses this issue by placing try/catches in high-level places to catch all possible exceptions. A helper message like the following is printed out after the exception's stack trace:

+---------------------------------------------------------+
|ERROR!!! An exception occurred while indexing.           |
|This is likely due to errors in database related things, |
|e.g. a mis-formed SQL query in the specification, a non- |
|existent column you specified, or the data isn't loaded  |
|into the database.                                       |
|                                                         |
|Indexing is now terminated and the server is restarted.  |
|Please inspect your spec and database, and then recompile|
|the project.If you can't figure out the issue, feel free |
|to contact Kyrix maintainers.                            |
|                                                         |
|Github: https://github.com/tracyhenry/kyrix              |
+---------------------------------------------------------+

Then, backend state is cleaned up and server is restarted. The user is then able to iteratively refine their spec without stopping docker at all.

Since #15 is also about checking SQL function/JS script securities (e.g. no DDL, DML), we'll leave it open.

This PR also has a fix to Kyrix-S which parses double values from strings more robustly.