- changed: execution is divided into parts by data type, because the execution time is significantly increased due to the greater number of possible options; splitting execution into separate parts also allows the test framework to run multiple parts in parallel

When generating code for a resource-constrained device, it is important to disable unused codecs and features. Unfortunately, the `asn1c` architecture does not allow you to simply use compiler/linker keys to exclude unused code, since references to unused code are always present in the generated files. Also, the need to compile even code that will not be used increases the build time.

- added: the following `asn1c` arguments to disable individual features:
    * `-no-gen-BER` (do not generate BER support code)
    * `-no-gen-XER` (do not generate XER support code)
    * `-no-gen-OER` (do not generate OER support code)
    * `-no-gen-UPER` (do not generate Unaligned PER support code)
    * `-no-gen-APER` (do not generate Aligned PER support code)
    * `-no-gen-print` (do not generate print code)
    * `-no-gen-random-fill` (do not generate random fill code)
- added: `asn_config.h` file containing the necessary definitions when disabling codecs and a feature
- changed: documentation and command-line help for added/modified arguments are updated
- check-parsing.sh: changed the parsing of `asn1c` arguments from the file name; now the list of arguments starts with the `+` sign; arguments are specified in the same way as when starting from the console (with the `-` sign at the beginning); arguments are separated by the `_` sign; example of the file name: `999-test-name.asn1.+-P_-gen-UPER_-gen-APER_-fcompound-names`
- fixed: file `160-multiple-parameterized-instance-OK.asn1.-Pgen-PERfcompound-names` empty because of an error parsing the arguments from the file name
- check-asn1c-smoke.sh (Important!): the execution time has significantly increased due to the greater number of possible options

* check if __EXTENSIONS__ if alreay
  defined

The encode function call will crash if it's not initialized. Update the document to make it more clear.

by Krzysztof Witek <kwitek@kw-net.com> vlm/asn1c d3dbe60400f80c91

Also, for the record. The following commits from vlm/asn1c PR #115 were
already suggested and merged here in the past:
- work in 128-bit integer values while compiling f49ee68bcb
- aper: use correct function for APER decoder    0abceb4382fee19c4
- aper: fix invalid read                         d158c573fb762857a

commit vlm/asn1c e9e346064b67a0946

Committed "aper: workaround for undefined enumeration indexes" by Krzysztof Witek <kwitek@kw-net.com>
commit a12c7fa47901bf6dee4cc21702efa6e7ecc1bb4f from vlm/asn1c #115

This fixes lots of typos that end up in the generated output, possibly
in other projects, which would then be unable to fix these typos as they
will reappear when running asn1c again.

Warned-by: codespell

Incomplete: Various fixes

* Empty fragment encoding
* Dereference DEFAULT to get the actual value
* Don't write length bits for small fixed-sized sequence-of
* Index encoding in choice extensions
* Only align when range == 256

converter-example.c.