Today I learned Chrome/Safari have an embedded gzip decompressor. So if you compress a FileAttachment with gzip -9 <filename> you can pretty easy decompressed it on the fly without any additional libraries. Can be a big saver.
I did it for bundling escodegen which required a few other annoying steps to get it to work on web but its a general technique I will continue to use.
Oh cool esm.run works (reliably?) but my point is the decompressionstream not the bundling of modules, this decompressionstream technique works very well for JSON datasets too where there is using a high quantity of shared structures (e.g. repeated keys in an array of objects).