translational recoding for UGA for Selenocysteine, UAG for Pyrrolysine, etc

[peterjc]

Raised by @idoerg on #1103, how can the existing Biopython translation function/method support translational recoding? e.g. UGA for Selenocysteine https://en.wikipedia.org/wiki/Selenocysteine
and UAG for Pyrrolysine https://en.wikipedia.org/wiki/Pyrrolysine

[peterjc]

In GenBank files this can be annotated with the /transl_except=... qualifier,

Quoting ftp://ftp.ncbi.nih.gov/genbank/gbrel.txt

/transl_except	Translational exception: single codon, the translation
		of which does not conform to the reference genetic code

Quoting https://www.ddbj.nig.ac.jp/FT/full_index.html

Qualifier       /transl_except=
Definition      translational exception: single codon the translation of which
                does not conform to genetic code defined by /organism or 
                /transl_table.
Value format    (pos:location,aa:<amino_acid>) where amino_acid is the
                amino acid coded by the codon at the base_range position
Example         /transl_except=(pos:213..215,aa:Trp)
                /transl_except=(pos:1017,aa:TERM)
                /transl_except=(pos:2000..2001,aa:TERM)
                /transl_except=(pos:X22222:15..17,aa:Ala)
Comment         if the amino acid is not on the restricted vocabulary list use
                e.g., '/transl_except=(pos:213..215,aa:OTHER)' with
                '/note="name of unusual amino acid"';
                for modified amino-acid selenocysteine use three letter code
                'Sec'  (one letter code 'U' in amino-acid sequence)
                /transl_except=(pos:1002..1004,aa:Sec);
                for partial termination codons where TAA stop codon is
                completed by the addition of 3' A residues to the mRNA
                either a single base_position or a base_range is used, e.g.
                if partial stop codon is a single base:
                /transl_except=(pos:1017,aa:TERM)
                if partial stop codon consists of two bases:
                /transl_except=(pos:2000..2001,aa:TERM) with
                '/note='stop codon completed by the addition of 3' A residues 
                to the mRNA'.

[peterjc]

Slightly related #1224 raises some more tricky translation corner cases, where a codon can be an amino acid or a stop codon, and this is documented at the genetic code table level.

[Artoria2e5]

My terrible, terrible idea is that we do both transl_except and table expansion. The former would be used for cases when the recoding is definitively known, the latter for messing about on hypothetical translations.

For transl_except, we need a new parameter on _translate_str encoding a simplified version, something like a [{"pos": [213,215], aa: "O"}, {"pos": [216,218], aa: "*"}].

For table, we pass a brand-new CodonTable object with modification, which seems already supported. Just copying fields from an existing table and changing them seems easy enough, at least for the common case of recoding a stop codon without ambiguity.