Improve Netzschleuder code

[krlmlr]

• Read in chunks using httr2::resp_stream_raw() : https://github.com/igraph/igraphdata/pull/23/files#r2047637905
• Use a named list as return for resolve_name(): https://github.com/igraph/igraphdata/pull/23/files#r2047636940
• Clean temporary file: https://github.com/igraph/igraphdata/pull/23/files#r2049479042
• Double-check use of [1] vs. [[1]] : https://github.com/igraph/igraphdata/pull/23/files#r2049490639
• Don't babysit the user, improve error message if needed: https://github.com/igraph/igraphdata/pull/23/files#r2049473272

[schochastics]

@krlmlr What is the advantage of resp_stream_raw()? Should this replace the current logic of checking filesize beforehand and only download if it doesn't go over a certain limit?

igraphdata/R/netzschleuder.R

Lines 57 to 64 in 0dcc3f3

	byte_size <- as.numeric(httr2::resp_headers(resp)[["content-length"]])
	gb_size <- round(byte_size / 1024^3, 4)
	if (gb_size > size_limit) {
	cli::cli_abort(c(
	"{zip_url} has a size of {gb_size} GB and exceeds the size limit of {size_limit} GB.",
	"i" = "To download the file, set {.arg size_limit} to a value greater than {gb_size}"
	))
	}

[krlmlr]

We can read in chunks of 1 MB or so, and are no longer limited regarding file size. The current code loads the entire file into RAM. I suspect this will also help with rate limiting, need to check.

Thinking about it, we need RAM proportional to the graph size anyway, so perhaps not important?