Various efficiency fixups for rfc1035 parser

None of these affect correctness, but do give us further
substantial reductions in the generated code size.  Combined with
the savings from the earlier parentheses preprocessing commit, the
size of the ragel-generated output is now ~67% smaller than in
earlier releases.

src/zscan_rfc1035.rl

@@ -340,7 +340,9 @@ static void text_add_tok(zscan_t* z, const unsigned len, const bool big_ok)
     unsigned newlen = len;
     if (len) {
         newlen = dns_unescape(text_temp, z->tstart, len);
-        gdnsd_assert(newlen && newlen <= len);
+        if (!newlen)
+            parse_error_noargs("Text chunk has bad escape sequence");
+        gdnsd_assert(newlen <= len);
     }
 
     if (newlen > 255U) {
@@ -397,7 +399,9 @@ static void text_add_tok_huge(zscan_t* z, const unsigned len)
     unsigned newlen = len;
     if (len) {
         newlen = dns_unescape(storage, z->tstart, len);
-        gdnsd_assert(newlen && newlen <= len);
+        if (!newlen)
+            parse_error_noargs("Text chunk has bad escape sequence");
+        gdnsd_assert(newlen <= len);
     }
 
     if (newlen > 16000U) {
@@ -419,6 +423,8 @@ static void set_filename(zscan_t* z, const unsigned len)
 {
     char* fn = malloc(len + 1);
     const unsigned newlen = dns_unescape(fn, z->tstart, len);
+    if (!newlen)
+        parse_error_noargs("Filename has bad escape sequence");
     gdnsd_assert(newlen <= len);
     z->include_filename = fn = realloc(fn, newlen + 1);
     fn[newlen] = 0;
@@ -584,7 +590,7 @@ F_NONNULL
 static void rec_mx(zscan_t* z)
 {
     validate_lhs_not_ooz(z);
-    if (ltree_add_rec_mx(z->zone, z->lhs_dname, z->rhs_dname, z->ttl, z->uv_1))
+    if (ltree_add_rec_mx(z->zone, z->lhs_dname, z->rhs_dname, z->ttl, z->uval))
         siglongjmp(z->jbuf, 1);
 }
 
@@ -655,8 +661,8 @@ static void rec_rfc3597(zscan_t* z)
 F_NONNULL
 static void rec_caa(zscan_t* z)
 {
-    if (z->uv_1 > 255)
-        parse_error("CAA flags byte value %u is >255", z->uv_1);
+    if (z->uval > 255)
+        parse_error("CAA flags byte value %u is >255", z->uval);
 
     validate_lhs_not_ooz(z);
 
@@ -667,7 +673,7 @@ static void rec_caa(zscan_t* z)
 
     uint8_t* caa_rdata = xmalloc(total_len);
     uint8_t* caa_write = caa_rdata;
-    *caa_write++ = z->uv_1;
+    *caa_write++ = z->uval;
     *caa_write++ = prop_len;
     memcpy(caa_write, z->caa_prop, prop_len);
     caa_write += prop_len;
@@ -895,8 +901,7 @@ static void preprocess_buf(zscan_t* z, char* buf, const size_t buflen)
 
     # Escape sequences in general for any character-string
     #  (domainname or TXT record rdata, etc)
-    escape_int = 25[0-5] | ( 2[0-4] | [01][0-9] ) [0-9] ;
-    escapes    = ('\\' [^0-9\n]) | ('\\' escape_int) | ('\\' nl);
+    escapes    = '\\' ( [^0-9\n] | [0-9]{3} | nl);
 
     # Quoted character string
     qword     = '"' ([^"\n\\]|escapes|nl)* '"';
@@ -947,10 +952,8 @@ static void preprocess_buf(zscan_t* z, char* buf, const size_t buflen)
     # IPv[46] Addresses.  Note that while they are not very
     #  very precise, anything bad that gets past them will still
     #  trigger graceful failure when passed to inet_pton().
-    ipoct     = digit{1,3};
-    _ipv4     = ipoct ('.' ipoct){3};
-    ipv4      = _ipv4 >token_start %set_ipv4;
-    ipv6      = ([a-fA-F0-9:]+ ( ':' _ipv4 )?) >token_start %set_ipv6;
+    ipv4      = [0-9.]+ >token_start %set_ipv4;
+    ipv6      = [a-fA-F0-9:.]+ >token_start %set_ipv6;
 
     # NAPTR's text strings
     naptr_txt = (txt_item_255 ws txt_item_255 ws txt_item_255) $1 %0 >start_txt;
@@ -962,8 +965,8 @@ static void preprocess_buf(zscan_t* z, char* buf, const size_t buflen)
     rfc3597_rdata = uval %set_uv_1 ws '\\' '#' ws uval %rfc3597_data_setup ws
         (rfc3597_octet+ ws?)**;
 
-    caa_prop = [0-9A-Za-z]{1,255} >token_start %set_caa_prop;
-    caa_rdata = uval %set_uv_1 ws caa_prop ws txt_item_huge >start_txt;
+    caa_prop = [0-9A-Za-z]+ >token_start %set_caa_prop;
+    caa_rdata = uval ws caa_prop ws txt_item_huge >start_txt;
 
     # The left half of a resource record, which for our purposes here
     #  is the optional domainname and/or the optional ttl and/or the
@@ -989,7 +992,7 @@ static void preprocess_buf(zscan_t* z, char* buf, const size_t buflen)
         | ('NS'i    ws dname_rhs) %rec_ns
         | ('CNAME'i ws dname_rhs) %rec_cname
         | ('PTR'i   ws dname_rhs) %rec_ptr
-        | ('MX'i    ws uval %set_uv_1 ws dname_rhs) %rec_mx
+        | ('MX'i    ws uval ws dname_rhs) %rec_mx
         | ('TXT'i   ws txt_rdata) %rec_txt
         | ('SRV'i   ws uval %set_uv_1 ws uval %set_uv_2
                     ws uval %set_uv_3 ws dname_rhs) %rec_srv