This data can be used to exercise every piece of code that works
with data, such as asn1 der generator and the asn1 codegen. More
generally it can be used to create random certificates for testing.
At the moment it generates random data but this could be extended
to generate edges cases as well.

Signed-off-by: Amaury Pouly <[email protected]>

Due to the hashmap using a nondeterministic order, it could differ
between runs. Sort the entries to make sure this does not happen.

Signed-off-by: Amaury Pouly <[email protected]>

The generator now output C++ unittest, those are compiled and
turned into a bazel test by the bazek certificate macro.

Signed-off-by: Amaury Pouly <[email protected]>

The code currently always allocates 3 bytes for the length and then
moves the data forward if we need less bytes than requires. This works
but means that the code could throw a kErrorAsn1BufferExhausted when
producing a document even though the final encoding would actually fit
into the buffer.

This commit changes the behaviour so that the code now allocates only
1 byte for the length initially and moves the data backwards on finish
if necessary. This guarantees that if the final document fits in the
buffer, then the code will successfully produces it. This requires a bit
of shuffling around in the code. It also requires a bit more care to
avoid an accidental buffer overflow when moving the data. There is a
unittest to specifically test corner cases in the moving code to make
sure we never write past the end of the buffer.

Signed-off-by: Amaury Pouly <[email protected]>

The code was double counting in some code path. Also the code was
under-estimating the size of unsigned integers due to a quirk of
ASN1 integer encoding that might require a padding with a 0x00
byte.

Signed-off-by: Amaury Pouly <[email protected]>

Signed-off-by: Amaury Pouly <[email protected]>

Signed-off-by: Amaury Pouly <[email protected]>

Signed-off-by: Amaury Pouly <[email protected]>

Signed-off-by: Amaury Pouly <[email protected]>