NASM is much more popular than TASM, but get_lexer_for_filename('.asm') would
return TASM. Resolved by changing the NASM priority to 1, and adding an
analyse_text implementation to both NasmLexer and TasmLexer to help resolve
which one to use (PROC should be fairly common in TASM code.)

Directives were parsed independent of whitespace after them, which caused the
cpuid instruction to be parsed as CPU & id, instead of cpuid. We now expect a
whitespace character after a directive, which seems to match the Nasm
documentation.

Add to CHANGES, languages, add versionadded, recreate mappings, fix DASM16
alias.

Lexers copied unmodified from
https://github.com/liluo/pygments-github-lexers
which is available under a 2-clause BSD license (same as pygments),
copyright 2012 to GitHub, Inc.

Fixes #1391 and #1150.

The keywords are based extracted from the source code
of the LLVM lexer (llvm-7.0.0.src/lib/AsmParser) and
is based on the first argument to the KEYWORD,
TYPEKEYWORD and INSTKEYWORD macros. The list of keywords
were then sorted in lexicographic order.

A diff with added and removed keywords is made available
at https://gist.github.com/mewmew/508e090a361095eeb60ffa88321e878a

Specifically the following 10 keywords were removed:

	begin, dbg, end, free, getresult, linker_private,
	linker_private, malloc, singlethread and x86_64_win64cc

And 96 new keywords were added.

Fix handling of end-of-line comments in asm lexer

Also, raise on warnings from Pygments only.

When there is a comment after an instruction, e.g.

     mov $1,%eax   ; move 1 to eax
     ret

the next instruction (ret above) is not properly highlighted. This
patch fixes that.

This was also reported in #1477.