Demonstrate lambda/excel issue

Minimal repo to demonstrate covidatlas/li#564

Currently, it appears that AWS lambdas (or sdk, or ... ?) mess around with what should be binary data. When I download an Excel file via lambdas (using arc), and I try to parse the file, I get an error:

x Error: End of data reached (data length = 10043, asked index = 347979759). Corrupted zip ?

Others have noted the same issue.

To repro:

• Clone this repo
• npm install everything
• Run node check-source.js -- this uses the xlsx npm package to parse the xlsx file in public
• Start the sandbox with npm run start
• Run ./do-crawl to run a crawler: it downloads the file from public using http/get-get-normal.
• Run node check-crawled.js, and you'll see the error:

.../crawled.xlsx.gz
wrote /path/to/crawled.xlsx
/path/to/node_modules/XLSX/jszip.js:272
            throw new Error("End of data reached (data length = " + this.length + ", asked index = " + (newIndex) + "). Corrupted zip ?");
            ^

Error: End of data reached (data length = 10043, asked index = 347979759). Corrupted zip ?

The associated lambdas are ... hopefully ... clear!