[Build] Rework Makefile rules to ease addition of LOC package

[gapisback]

This commit refactors the Makefile rules to ease the addition of rules to integrate Line-of-Code (LOC) package in a follow-on commit. Several redundant TEST_BIN etc. symbols are purged.

Existing capabilities are unchanged, and we now additionally support:

$ make clean && CC=g++ CXX=g++ LD=g++ make all
$ make run-tests

... to build all .c, .cpp and .cc sample-programs and run them in one-pass.

Added test.sh wrapper script for devs to quickly re-run all the tests that are run in CI, for easily stabilizing changes w/o having to go thru CI.

---

Updated make help usage messages (as a quick reference):

agurajada-Linux-Vm:[1559] $ make help

Usage:

To build all sample programs and run all unit-tests:
 make clean && CC=g++ CXX=g++ LD=g++ make all
 make run-tests

Alternatively, you may use multiple invocations in this sequence to build all sources:

To build C-sample programs and run unit-tests:
 make clean && CC=gcc LD=g++ make all-c-tests
 make run-c-tests

To build C++-sample .cpp programs and run unit-tests:
 make clean-l3 && CC=g++ CXX=g++ LD=g++ make all-cpp-tests
 make run-cpp-tests

To build C++-sample .cc programs and run unit-tests:
 make clean-l3 && CC=g++ CXX=g++ LD=g++ make all-cc-tests
 make run-cc-tests

Environment variables:
 BUILD_MODE={release,debug}
 BUILD_VERBOSE={0,1}

NOTE: To the reviewer:

This is really a non-functional, code-cleanup, refactoring change.

I have folded-in the support for LOC package on top of this code in a test PR #8 , which shows that this refactoring greatly simplifies adding in Makefile rules to pick-up LOC's generated files and other Make-rules.

Additionally, in this change-set, you will see LOC making an appearance in l3.h and l3.c files:

I have edited the files to "prepare" the structures and L3 APIs to start working with 4-byte loc_t ID value.

Currently, all this is #ifdef'ed out under L3_LOC_ENABLED define.

In a future PR, the builds will be invoked with this -DL3_LOC_ENABLED which will - magically - activate this code, and L3 logging will become LOC-enabled.

Once this goes in, I am ready to submit a PR for integrating LOC with L3 package.

[gapisback]

No new tests added to CI. Just adding blank lines for readability.

[gapisback]

New functionality being onlined here.

[gapisback]

All these TEST* symbols on L137 were found unnecessary, so I've simplified them.

[gapisback]

This whole deleted block is now unnecessary as the dependencies of make targets on sample program binaries is defined differently, above.

[gapisback]

Here onwards, in the change-set, you will see LOC making an appearance.

I have edited the files to "prepare" the structures and L3 APIs to start working with 4-byte loc_t ID value.

Currently, all this is #ifdef'ed out under L3_LOC_ENABLED define.

In a future PR, the builds will be invoked with this -DL3_LOC_ENABLED which will - magically - activate this code, and L3 logging will become LOC-enabled.

[gapisback]

In an immediately upcoming PR #8 , if we build the sample programs with L3_LOC_ENABLED=1, then we will store the LOC-ID token in the L3 logging entry struct. Otherwise, will just store 0.

[gapisback]

Note that in a future PR, when this field is activated, we are -only- re-using the 4-byte hole that is conveniently left behind for us in current non-L3_LOC_ENABLED build:

(gdb) ptype/o struct l3_entry
/* offset      |    size */  type = struct l3_entry {
/*      0      |       4 */    pid_t tid;
/* XXX  4-byte hole      */                                <<< loc_t loc; field is using these pad bytes.
/*      8      |       8 */    const char *msg;
/*     16      |       8 */    uint64_t arg1;
/*     24      |       8 */    uint64_t arg2;

                               /* total size (bytes):   32 */
                             }